Qwen4-Max

(qwen.ai)

731 points by ilreb13 hours ago

37 comments

  • simonw11 hours ago
    Qwen4-Max pelicans at four reasoning efforts (none, low, high, xhigh): <a href="https:&#x2F;&#x2F;tools.simonwillison.net&#x2F;markdown-svg-renderer?url=https%3A%2F%2Fgist.github.com%2Fsimonw%2Fff451620c173ea21e21b97b8cd46805b" rel="nofollow">https:&#x2F;&#x2F;tools.simonwillison.net&#x2F;markdown-svg-renderer?url=ht...</a><p>For comparison, here&#x27;s one I got from Qwen3.8 Max (at default reasoning effort): <a href="https:&#x2F;&#x2F;simonwillison.net&#x2F;2026&#x2F;Jul&#x2F;20&#x2F;afraid-of-chinese-models&#x2F;" rel="nofollow">https:&#x2F;&#x2F;simonwillison.net&#x2F;2026&#x2F;Jul&#x2F;20&#x2F;afraid-of-chinese-mode...</a><p>This inspired me to re-run pelicans through 3.8 Max - but OpenRouter has qwen3.8-max-prime and qwen3.8-max-0902 - prime is twice the price and faster - so I went with the 0902 one.<p>Turns out Qwen4-Max has a none setting, but 3.8 Max didn&#x27;t allow disabling reasoning - you had to pick from these five levels instead: minimal, low, medium, high, xhigh.<p>Here are those 3.8 Max 0902 pelicans: <a href="https:&#x2F;&#x2F;tools.simonwillison.net&#x2F;markdown-svg-renderer?url=https%3A%2F%2Fgist.github.com%2Fsimonw%2F88071abe6ecd0f8e13ab77ade259be9e" rel="nofollow">https:&#x2F;&#x2F;tools.simonwillison.net&#x2F;markdown-svg-renderer?url=ht...</a> - they&#x27;re mostly not as good as 4, but I prefer the 3.8 xhigh one.
    • dtech8 hours ago
      I don&#x27;t fully understand why you keep doing these as-is. You original motivation was that this was not in the training set and indeed it was a great benchmark for a while, but it&#x27;s clearly been incorporated now in the training data and possibly internal benchmarks.
      • trentor4 hours ago
        Because he wants the engagement they generate. I don&#x27;t understand why they still get upvoted on HN, but here we are. If even I can recognize that they&#x27;re pretty much useless as benchmarks at this point, he must know it too.
        • QuantumNomad_4 hours ago
          He’s our pelican man :) I will always upvote pelican posts from the HN resident pelican guy
          • doe881 hour ago
            Likewise, i always upvote these posts even if i know it is biaised, it is always fun to see their evolutions, there is nothing wrong with them. Also even if we know there are trained on them specifically, some models are nonetheless better than others in engineering the output, thus it is always interesting.
          • dvfjsdhgfv54 minutes ago
            You see it was funny the first and second time two years ago or so, now it&#x27;s just boring. Whenever I want to read some insightful comments about a new model I need to scroll down through a few pages of comments about pelicans as if they meant anything now.
            • nickalaso9 minutes ago
              You can click the &#x27;-&#x27;&#x2F;&#x27;next&#x27; symbol on a parent comment to collapse it. You are welcome :)
            • bee_rider34 minutes ago
              He seems to have posted it as a top-level comment, so it should be possible to fold it all, right? If he hadn’t, multiple other people might be posting about pelicans. Consider it the pelican attracting sacrifice post.
          • embedding-shape4 hours ago
            Things like &quot;I will always upvote X posts from the HN resident X&quot; feels like what HN should actively work against.
            • GTP3 hours ago
              I think their goal was just to write a funny comment.
              • CryptoBanker1 hour ago
                Which is also something we try to avoid here
              • embedding-shape1 hour ago
                Yeah, which I suppose used to be something we all kind of tried to avoid here, to make conversations more curious instead of just pilling on with memes or whatever other communities been degraded to nowadays.
                • GTP3 minutes ago
                  In my view, as long as it is just an occasional funny comment here and there, it is not an issue.
          • Betelbuddy4 hours ago
            I have a colleague who got to the last rounds of interviews at a FANG, but did not make it...He could not draw a Pelican...
            • GTP3 hours ago
              That&#x27;s why AI will replace us: will be able to draw better pelicans:D
        • micoti2 hours ago
          It&#x27;s shame that people try to farm engagement on HN. Is this transforming to reddit. I don&#x27;t know at this point but I agree with parent comment. EDIT: One thing I know is that I now need to scroll past the pelican thread if I need to read something usefull
        • latentsea40 minutes ago
          It&#x27;s part of culture at this point.
        • nickalaso10 minutes ago
          Based on some of the trash that consistently makes it to the front page I am very much ok with Pelican Man posting comments linking to his benchmark, even if its a bit less meaningful than it used to be.
        • keyle4 hours ago
          And he spends the time doing it and as part of it drops nuggets of information not seen anywhere else. Respectfully and highly welcome. You&#x27;re most welcome to move on.
      • powersnail2 hours ago
        Look at the page, though. If the data is in their training set, then the training must not have been very good, cause the images have very salient defects, and random artifacts popping up.<p>As far as I&#x27;ve seen, among the popular models today, there are still significant gaps in the quality of bike-riding pelikan svgs, making the benchmark still useful. Maybe not as a differentiator of frontier models, but certainly still informative for lesser competitors.
        • paimapi49 minutes ago
          Unless the website is pulling hundreds of thousands of views per pelican, I sort of doubt any firm is actually training or post-training to optimize for this benchmark. They&#x27;ll more generally optimize for better SVG rendering and I imagine these pelicans arose out of the annoyance we all had many months ago with how utterly bumbling and forgetful models used to be when completing tasks like that.
      • andy_ppp6 hours ago
        He can do whatever he wants, I personally think they are quite informative, albeit I don’t trust them that much either. It’s like a vague heuristic to say “this looks like an interesting model to test out”.
        • trio84535 hours ago
          Is it informative? I really don&#x27;t get what signal you&#x27;re picking out of a bunch of similar looking drawings.
          • anentropic5 hours ago
            It&#x27;s interesting to see which model + thinking strength gets the details right, such as legs on opposite sides of the frame, a coherent bicycle etc
            • andy_ppp4 hours ago
              Some people are determined not to understand I think, but if they don&#x27;t find it useful no worries, there&#x27;s just no need to have a temper tantrum about what other people post on here especially when the majority (it&#x27;s near the top for a reason) find it useful. Just don&#x27;t worry about it - you can even hide whole threads on here if you don&#x27;t like them, I use the feature all the time rather than taking the time to comment negatively!
            • skohan2 hours ago
              Isn&#x27;t it a bit arbitrary though?<p>The original concept of this &quot;benchmark&quot; was to see how the model performs on a reasonably complex task that probably isn&#x27;t in the training data.<p>Even if they don&#x27;t fine-tune for this specifically, there are now surely tons of examples of people asking for a pelican on a bike in the training data, and making an SVG isn&#x27;t that complex of a task for current gen models.<p>Even with the same model at the same thinking level, you might get different quality out of the same prompt on different runs. If you want to use this as an eval, you would really want to generate 100 of them, and see how often they&#x27;re anatomically&#x2F;mechanically accurate.
        • CodeNest6 hours ago
          [dead]
      • raincole6 hours ago
        It&#x27;s more a meme&#x2F;influencer thing now.
        • Jtarii4 hours ago
          Just the hacker news equivalent of when does the narwhal bacon at this point.
      • cheema338 hours ago
        &gt; but it&#x27;s clearly been incorporated now in the training data<p>You are not the first one to think of this. And Simon has thought of it as well, and commented on it several times.
        • KeplerBoy8 hours ago
          Still, by now it&#x27;s just more tradition than anything else. It&#x27;s just kind of neat to have a consistent benchmark across many LLMs, even if the benchmark is completely gamed, but that is the nature of benchmarks.
      • kalcode2 hours ago
        Well IMO everyone can make their own benchmarks pretty easily, and that&#x27;s what we should be doing.<p>Harder to train off all them.<p>Aka, let him have his fun, others are making benchmarks too
      • MrScruff2 hours ago
        If this is that case it should be easy to demonstrate by asking current models for an svg of a rhino surfing. My guess is you would see a similar trend, given that the latest models are demonstrating large leaps in all sorts of content creation including 3D animation, rather than this being a demonstration of some sort of pelican-maxxing.
      • ciefa6 hours ago
        I love these. They&#x27;re fun.
        • ErneX1 hour ago
          Agree.
        • yokoprime1 hour ago
          I absolutely agree. its also fun to see the fun-hating gremlins emerge every time simon posts these.
        • InsideOutSanta5 hours ago
          OK, but why do you like fun, though?
          • tacet5 hours ago
            fun? how dare you?!
      • IsTom8 hours ago
        They might be in training set, but there&#x27;s still things wrong in them. That&#x27;s not a great look for these companies IMO.
      • wongarsu5 hours ago
        It keeps getting upvoted, so clearly a lot of HN users still find it valuable
      • harveymadeunico3 hours ago
        Because it’s fun! The moment pelicans stop is a sad day. Keep them coming!
      • swader9992 hours ago
        How can they not keep doing it? It&#x27;s become unstoppable.
      • blitzar6 hours ago
        I feel bad for the data scientist that has to find or create images of pelicans riding bicycles for the training set to pass the meme benchmark.<p>Im not sure how you get the pelican to sit still let alone pedal, but there is a reason ai salaries are so high.
        • irthomasthomas6 hours ago
          They hire professionals and subject matter experts to produce the training data.
          • codeulike5 hours ago
            So they hire a professional pelecanus velocipede ideator?
            • GTP2 hours ago
              Don&#x27;t know if they do, but they could hire a cartoonist.
            • reylas1 hour ago
              This makes more sense if you pretend it is Dr Doofenshmirtz.
      • eddyg3 hours ago
        I wouldn’t be too sure.⁽¹⁾<p><i>“there’s little evidence that AI labs are pelicanmaxxing. Or at least they’re not doing it in a plainly obvious manner. Pelicans aren’t drawn any better than other animals. Bicycles aren’t drawn any better than other vehicles. And no lab draws the combination better than its pelicans and bicycles already predict.”</i><p>⁽¹⁾ <a href="https:&#x2F;&#x2F;dylancastillo.co&#x2F;posts&#x2F;pelicanmaxxing.html" rel="nofollow">https:&#x2F;&#x2F;dylancastillo.co&#x2F;posts&#x2F;pelicanmaxxing.html</a>
      • hmokiguess2 hours ago
        I feel like there&#x27;s emotional attachment and influencer branding energy with this by now. It&#x27;s more like a developer relations or developer ambassador type of content.
      • roenxi7 hours ago
        &gt; ...but it&#x27;s clearly been incorporated now in the training data...<p>It clearly hasn&#x27;t; if it was quality of the pelicans and bikes would be static over time. Instead the quality has been improving along with the models.
      • 0-_-07 hours ago
        Because it gets upvoted which means there is demand for it here
      • Betelbuddy4 hours ago
        He should keep doing Pelicans, but every time throw in a surprise animal and a surprise locomotion vehicle. Would help identify the ones training for the Pelican.<p>Would love to see an SVG of an Elephant on a skateboard...
      • jMyles1 hour ago
        Someone, somewhere, at some AI lab, has by now said something like:<p>&quot;We need to make the pelicans look good, but not so good that it looks like we&#x27;re using them as an internal benchmark.&quot;
      • alienbaby5 hours ago
        It&#x27;s a little light hearted fun?
      • fortyseven2 hours ago
        It&#x27;s a fun tradition. Remember fun? We used to have that on the fucking internet.
        • sirsinsalot1 hour ago
          I mostly remember flame wars about editors and limewire
      • yokoprime1 hour ago
        Simon has clearly stated previously that he compares with other &quot;animal doing human thing&quot; type prompts to compare and see if theres a chance they are cheating by training on pelican svgs. Thus far, nobody has been caught cheating.
        • geektips1 hour ago
          Actually, a Google employee has publicly mentioned that they’re specifically working on improving SVG generation in Gemini. Simon Willison has also pointed this out when evaluating Gemini models. [^1]<p>[^1] <a href="https:&#x2F;&#x2F;twitter.com&#x2F;sunjiao123sun_&#x2F;status&#x2F;202455551655137292" rel="nofollow">https:&#x2F;&#x2F;twitter.com&#x2F;sunjiao123sun_&#x2F;status&#x2F;202455551655137292</a>
      • lexh7 hours ago
        I more don&#x27;t understand why these comments get upvoted to top comment all the time.<p>Like... we get it. Simon likes pelicans. If I want to find his boondoggle robo pelican drawings, I know where to find them by now.<p>No need to spam every single model release thread.
        • mirekrusin7 hours ago
          You can click [-] to collapse thread.
          • lexh5 hours ago
            Yeah fair play. I was just trying to add a different angle to OP&#x27;s grievance. Perhaps it&#x27;s time for a browser extension to hide these for me.
        • windexh8er3 hours ago
          Agreed. It&#x27;s really annoying to come into the thread and a link to a pelican I have no interest in ever seeing is the top comment. Does anyone think these have any actual value any more?
          • bitexploder3 hours ago
            Yes, if you pay close attention to what they infer. Just click [-] and move on.
            • windexh8er2 hours ago
              I&#x27;m curious. What&#x27;s the actual value, then? I fully understand minimizing a thread, but also am free to voice my opinion the same as you.<p>So, again... What should I be &quot;paying close attention&quot; to?
              • bitexploder1 hour ago
                Alright, I was rude. So.. you can’t really benchmaxx pelicans specifically. You have to train the whole model on SVG and that is just generally useful. Secondarily it shows how a model understands the objects themselves and how to express it abstractly. IDK, it is a lot more interesting than assumed.<p>Here are a bunch of Pelicans I made in my basement. Side by side quant comparisons and something for fun :)<p><a href="https:&#x2F;&#x2F;stealthgoat.com&#x2F;fnfl&#x2F;pelicans3.html" rel="nofollow">https:&#x2F;&#x2F;stealthgoat.com&#x2F;fnfl&#x2F;pelicans3.html</a> <a href="https:&#x2F;&#x2F;stealthgoat.com&#x2F;fnfl&#x2F;pelicans8.html" rel="nofollow">https:&#x2F;&#x2F;stealthgoat.com&#x2F;fnfl&#x2F;pelicans8.html</a> My favorite, and shows this is not all benchmaxxing, this model is just good at SVG: <a href="https:&#x2F;&#x2F;stealthgoat.com&#x2F;fnfl&#x2F;pelicans6.html" rel="nofollow">https:&#x2F;&#x2F;stealthgoat.com&#x2F;fnfl&#x2F;pelicans6.html</a> Oh, and, quantization matters: <a href="https:&#x2F;&#x2F;stealthgoat.com&#x2F;fnfl&#x2F;pelicans2.html" rel="nofollow">https:&#x2F;&#x2F;stealthgoat.com&#x2F;fnfl&#x2F;pelicans2.html</a><p>Heres some examples.
        • ciefa6 hours ago
          Man, some of you people really have no whimsy :D
        • drcongo6 hours ago
          On the flip side, I love the pelican posts, but I find the people who think they&#x27;re cleverly pointing out for the very first time that the benchmark may have been gamed by now tedious in the extreme.
          • latentsea35 minutes ago
            Then perhaps I should point out something different. Namely, I suspect this will thoroughly confuse the machines and instead of optimizing the universe to generate paperclips they will instead optimize it to generate pelicans.
          • irthomasthomas6 hours ago
            It&#x27;s september
    • nickalaso12 minutes ago
      At this point Simon can we maybe think about making this a 3d model? Maybe GTLF&#x2F;GLB or custom Json you could even keep closed source for some kind of barrier to it getting trained on?<p>I say this because, even just based on my own product related to cad design&#x2F;3d modeling, the SOTA models have gotten very good at 3d and I don&#x27;t think the 2d SVG generation tells us much anymore. Maybe I can whip this up myself on my own site if anyone is interested tbh.
    • irthomasthomas3 hours ago
      If you use the same prompt via the qwenapp or qwen studio you get an interactive web page with an animated scene. You can control every element of the scene, the colors, speed, time of day l, visible accessories etc and then download the one you like best.<p><a href="https:&#x2F;&#x2F;chat.qwen.ai&#x2F;s&#x2F;deploy&#x2F;t_5fb28dd7-a61f-4946-8e59-8dd1c43d015d" rel="nofollow">https:&#x2F;&#x2F;chat.qwen.ai&#x2F;s&#x2F;deploy&#x2F;t_5fb28dd7-a61f-4946-8e59-8dd1...</a>
    • Anon109611 hours ago
      I feel like the xhigh result is one of the first times I&#x27;ve seen a model go for a non-red colored bicycle (teal). Still haven&#x27;t seen the leftward facing bike though.
      • bitexploder3 hours ago
        Teal is a popular second color for Qwen Flash Next 3.8 as well. Also, I am not sure people realize how much variance there is for the same prompt. Enjoy.<p><a href="https:&#x2F;&#x2F;stealthgoat.com&#x2F;fnfl&#x2F;pelicans3.html" rel="nofollow">https:&#x2F;&#x2F;stealthgoat.com&#x2F;fnfl&#x2F;pelicans3.html</a><p><a href="https:&#x2F;&#x2F;stealthgoat.com&#x2F;fnfl&#x2F;pelicans8.html" rel="nofollow">https:&#x2F;&#x2F;stealthgoat.com&#x2F;fnfl&#x2F;pelicans8.html</a><p>My favorite, and shows this is not all benchmaxxing, this model is just good at SVG: <a href="https:&#x2F;&#x2F;stealthgoat.com&#x2F;fnfl&#x2F;pelicans6.html" rel="nofollow">https:&#x2F;&#x2F;stealthgoat.com&#x2F;fnfl&#x2F;pelicans6.html</a><p>Oh, and, quantization matters: <a href="https:&#x2F;&#x2F;stealthgoat.com&#x2F;fnfl&#x2F;pelicans2.html" rel="nofollow">https:&#x2F;&#x2F;stealthgoat.com&#x2F;fnfl&#x2F;pelicans2.html</a>
      • inshard11 hours ago
        Astra and the OpenAI models do this color as well
        • gpt511 hours ago
          Evidence from where they distilling from &#x2F;s
    • Starlevel00411 hours ago
      Feels like Pelicanbench is getting saturated now. Maybe time to switch to a different animal and vehicle combo?
      • wasabi99101110 hours ago
        You say that, but it still takes quite a bit of reasoning to get the pelican&#x27;s legs on different sides of the bike and to get the bike chain included.
        • Melatonic8 hours ago
          I feel like randomly even at max reasoning they get the legs wrong. Or the pedal. Or both.
          • msandford2 hours ago
            I don&#x27;t know if I&#x27;ve ever seen a bent leg. Legs always go straight to pedals with no knees. The models might have an idea of what it looks like but seem to have no understanding of how it works.
      • slashtrends11 hours ago
        The 3d one still produces interesting variations - <a href="https:&#x2F;&#x2F;www.reddit.com&#x2F;r&#x2F;ClaudeCode&#x2F;comments&#x2F;1xs4ll7&#x2F;fable_52_vs_opus_vs_astra_pelican_riding_a&#x2F;" rel="nofollow">https:&#x2F;&#x2F;www.reddit.com&#x2F;r&#x2F;ClaudeCode&#x2F;comments&#x2F;1xs4ll7&#x2F;fable_5...</a>
      • numpad07 hours ago
        Or something else entirely, but you&#x27;ll still need the pelican for comparison, I would think.<p><a href="https:&#x2F;&#x2F;aishodo.enpitsu.org&#x2F;?lang=en" rel="nofollow">https:&#x2F;&#x2F;aishodo.enpitsu.org&#x2F;?lang=en</a>
      • dmit3 hours ago
        Pogo-stick jellyfish, let’s go
      • sorrywhat10 hours ago
        First thing I do when a new model drop is scroll down to find the pelican comment, and then the &quot;the pelican benchmark is no longer useful&quot; debate under it.
      • Melatonic8 hours ago
        What if it&#x27;s been a Stork this whole time?
      • forrestthewoods9 hours ago
        Kind of yes. Except from an artistic perspective they&#x27;re all still awful.
    • lcnPylGDnU4H9OF11 hours ago
      I kinda like the no-effort result more than low, but high and xhigh return to what one would expect for reasoning-&gt;quality.
    • bergheim2 hours ago
      I don&#x27;t understand why this is always the most upvoted thing for a new model. It&#x27;s like some influencer &quot;data&quot; at this point. We&#x27;re basically in 2027 now, the pelicans had their run, it&#x27;s over.<p>Not a dig at you Simon, you&#x27;ve done good stuff, more just in general. Why is this always upvoted, it&#x27;s like reading The Sun or something.<p>And why always one image. Should it not be say a 100 images then averaged or something? A prompt with temperature is a slot machine in Vegas, wtf does that have to do with model quality.
      • dd8601fn2 hours ago
        It’s a fun little exercise where the output roughly correlated with the progress of models.<p>The “real” benchmarks are widely available, and frankly kinda sketchy anyway, so why not have a little fun?
    • lobo_tuerto1 hour ago
      It&#x27;s time to move from pelicans to horses.
    • losthobbies5 hours ago
      You should call this The Pelican Brief
    • gcanyon2 hours ago
      Man, how hard is it to understand that you straddle a bike, and one leg should be behind the bike? Apparently pretty hard, since all but the xhigh fail this ¯\_(ツ)_&#x2F;¯
    • ReptileMan5 hours ago
      It has the legs on the proper sides of the frame only on max reasoning, which by coincidence was also the case with opus.<p>So either perspective is properly hard or the moonshine is strong in this one.
    • kronks1 hour ago
      [dead]
  • maplet12 hours ago
    It scores a 43 on Artificial Analysis with an average cost of $3.27 [1]. For reference, GPT-6.1 Sol (medium) gets a 46 with an average of $0.36.<p>[1]: <a href="https:&#x2F;&#x2F;artificialanalysis.ai&#x2F;models&#x2F;qwen4-max" rel="nofollow">https:&#x2F;&#x2F;artificialanalysis.ai&#x2F;models&#x2F;qwen4-max</a>
    • irthomasthomas5 hours ago
      We should stop using this site. They now include private benchmarks which cannot be replicated. And every time an openai model falls down they change the index until it floats to the top again.
      • davedx3 minutes ago
        Terminal Bench is great, public evals, very complete and highly complex. I was browsing through them today and it gave me a real appreciation for the benchmarkers.
      • rhdunn5 hours ago
        The key issue with public benchmarks is that they allow the model providers to train against them and artificially inflate their score accordingly. Private benchmarks are a way of making it harder to game the system, though it makes it harder for people looking at the benchmark to assess it as well.<p>Neither option is perfect.<p>A third option would be updating the benchmark every 6-12 months, but that has its challenges as well as you need to make sure that the benchmark covers things the model hasn&#x27;t seen or is harder for the models to game.
        • bee_rider21 minutes ago
          It’s also worth keeping in mind the other alternative, having internal benchmarks that more accurately reflect your workload, that you know all about, and the LLM provider doesn’t. There doesn’t seem to be a way to maintain such a thing specifically for the purposes of online chatter (both because it isn’t profitable, and because things like “your workload” aren’t possible to define in that case). So we should be happy with what we’ve got.<p>This is a permanent problem in benchmarking really. Transparency is required for trust but invites over-tuning.
          • rhdunn5 minutes ago
            There are tools like promptfoo that can be used for internal benchmarks and evaluations that allow you to compare results across different models and runs. This can be useful for personalized benchmarks and for business applications where you need to evaluate models and&#x2F;or system prompts.
        • aenis5 hours ago
          Makes me wonder how do you privately benchmark a model hosted by the lab. Even with zdr, the bytes are flowing through their infra and possibly influencing god knows what for the next training round.
          • nsagent4 hours ago
            You use something like item response theory [1] to be able to change the test over time while maintaining validity. It&#x27;s what standardized tests like the GRE use.<p>Interestingly, Artificial Analysis does have a score that uses IRT, it&#x27;s their Progress Index [2].<p>[1]: <a href="https:&#x2F;&#x2F;en.wikipedia.org&#x2F;wiki&#x2F;Item_response_theory" rel="nofollow">https:&#x2F;&#x2F;en.wikipedia.org&#x2F;wiki&#x2F;Item_response_theory</a><p>[2]: <a href="https:&#x2F;&#x2F;artificialanalysis.ai&#x2F;methodology&#x2F;progress-index" rel="nofollow">https:&#x2F;&#x2F;artificialanalysis.ai&#x2F;methodology&#x2F;progress-index</a>
          • runeblaze23 minutes ago
            this is an worry I sometimes see. I want to note that these are billion&#x2F;trillion dollar companies we are talking about that have mutual talent movement (= uncontrolled information flow). Violating ZDR might be contractual violation, or at the very least, a strong breach of trust.<p>(not to say there is no value in &quot;zero-trust&quot; environments, but often pure-benchmaxxing is a deadend; some benchmaxxing is always involved and much less risky options than &quot;train on the ZDR test env&quot; usually exist)
          • loehnsberg1 hour ago
            Would OpenAI also &quot;see&quot; the request if it went through something like AWS Bedrock? This isn’t their infra and afaik the whole idea of this kind of intermediary.
            • staticman242 minutes ago
              I thought the major benchmark companies always get advanced access to a model, which presumably is not done via bedrock.
        • irthomasthomas5 hours ago
          The private benchmarks can be seen by the labs when you run them. We only have their word that they wont look at them, which isn&#x27;t worth much. We should assume that everyone benchmaxxes equally, apart from a few outliers, and try to climb the hill by producing relevant training data.
          • rhdunn4 hours ago
            Yes, that&#x27;s another challenge with constructing benchmarks.<p>You could run the benchmarks using downstream providers like AWS if they are available.<p>You could vary the parameters of the tests to both help with LLMs learning as is and to help obscure that this is a test. Especially if sending them through a thirdparty, so the providers don&#x27;t get identifying markers like IP addresses.<p>But ultimately it&#x27;s another arms race like with viruses and anti-virus software.
          • Topfi2 hours ago
            Eh, I have an eval suite dating back to before GPT-5 in parts and have yet to see certain labs do as well on that as the public eval data would suggest. Course, I am pretty small, this is anecdotal and n=1, though I have seen very small, niche evals being clearly trained on the second they became public on GitHub for what that’s worth.
            • paimapi41 minutes ago
              It depends entirely on the financial incentive. Does improving against your eval suite lead to more adoption? Or would focusing on the far more popular benchmarks convince more CTOs to sign a multi-year contract?
              • Topfi18 minutes ago
                Who&#x27;d sign a multi-year contract based on any benchmarks, public, private, mine or someone elses? We get new frontier LLMs every 0.5-2 months and any CTO of a major entity already uses Bedrock or the like, so why would anyone lock themselves into a provider or even worse a single model for multiple years?
        • skohan2 hours ago
          I think the best solution is really to update the benchmark periodically, and publish the benchmarks after the results are published.<p>Otherwise you have to rely completely on trust that AA is really measuring intelligence.
        • Uke3 hours ago
          Or to separate the benchmark data by type and enable the user to filter by private, public or both.
        • tallesborges923 hours ago
          Livebench does it
        • Eisenstein5 hours ago
          A fourth option is realizing that benchmarks exist because it is impossible to measure the value we want to measure, so we need a stand-in. The stand-in will sometimes correlate with performance but often just act as a way to compare. When used to compare, it is most important that the benchmark be fair and trustworthy.
      • Topfi2 hours ago
        I have to defend private benchmarks on this front, big believer in them, have seen how they can surface major performance differences.<p>Certain labs have been shown to and will continue to train on public evals, whether unintentional or on purpose, any respectable ranking across multiple evals also needs private data to counteract this.
      • szambetti5 hours ago
        We really should. I&#x27;ve been using other benchmarks like <a href="https:&#x2F;&#x2F;livebench.ai&#x2F;#&#x2F;?cats=Coding" rel="nofollow">https:&#x2F;&#x2F;livebench.ai&#x2F;#&#x2F;?cats=Coding</a> and it is more aligned to what it feels to my experience as well
        • CharlieDigital2 hours ago
          This has Terra 6.1 at 3rd place on coding.<p>I have been using Terra a lot and I&#x27;d say that most of the models (even small ones) are are roughly equivalent when it comes to working in a consistent, established codebase.<p>But in practice, Terra is not very good at staying on task compared to say Sol 6.1 or Sonnet 5.5. Terra 6.1 will often just pause, thinking the objective has been met or the turn has ended.
      • fancyfredbot3 hours ago
        I don&#x27;t understand your point about openai. I can&#x27;t remember the last time openAI was at the top of the leaderboard? Google had a brief moment at the top recently but other than that I feel like it&#x27;s always Anthropic.<p>I kind of agree that closed benchmarks are bad. With a closed benchmark you still send the questions to the model provider and they may well train on it. Might as well publish it.
        • irthomasthomas2 hours ago
          Not the very top, but Astra was behind Kimi K3.1 before AA&#x27;s last change which boosted openai to second place. This has happened with every AA update.
      • andai1 hour ago
        Even worse, it lags on my device!<p>Do you know a better site?
      • Art96811 hour ago
        If they made the benchmarks public then Qwen would train on those to give the appearance their models are close to the American frontier. They have never been close. It&#x27;s well established Chinese companies cheat the benchmarks which is why until recently they gave the appearance of being close to frontier but in actual real use they are far behind.<p>You can easily come up with a difficult private benchmark and test them yourself to validate the intelligence gap. It is real.<p>China lost. There is no catching up. Ever. That ship sailed.
      • bluebands2 hours ago
        private benchmarks are good
    • esperent12 hours ago
      The table on that page shows gpt 6.1 sol (max) at $1.27 per task and it&#x27;s the only model cheaper than Qwen 4 max.<p>By comparison opus 5.5 (max) is $10.01.<p>If they are not testing Chinese models with less than max thinking, then we should only compare them to other models also at max thinking.
      • athrowaway3z9 hours ago
        &gt; If they are not testing Chinese models with less than max thinking, then we should only compare them to other models also at max thinking.<p>This is wrong.<p>To make a vacuum analogy, one brand&#x27;s Low setting could be another&#x27;s Medium setting.<p>To compare you either give them the same resources ( Watts or $ ) and see how much they can do, or you give them a task and then check how many resources they required to finish it.<p>The label on the knob is irrelevant, except that either model might have a setting higher than the other which is sometimes useful if the resources aren&#x27;t a constraint or lower when they are.
        • esperent6 hours ago
          &gt; To make a vacuum analogy, one brand&#x27;s Low setting could be another&#x27;s Medium setting<p>Which is why you need to test a range of settings.<p>Suppose you were testing car mileage. For the big name popular cars, you do 10km drive at 20km&#x2F;hr, 40km&#x2F;hr, 60km&#x2F;hr and then publish the results.<p>For all the less popular cars you only do one test at 60km&#x2F;hr and then you publish all the results for all cars on one big &quot;Pareto frontier&quot; graph claiming to show which car will allow you to drive home most efficiently.<p>It happens that lots of the less popular cars are not very efficient at 60km&#x2F;hr so naturally the graph makes them look bad. Would they also be bad if tested at 20km&#x2F;hr or 40km&#x2F;hr?<p><i>Nobody knows</i>. Because this is where the analogy collapses - unlike cars, hardly anyone is doing these kind of tests for LLMs.
          • spider-mario1 hour ago
            Well, sure, it would be better to evaluate all of them throughout the range. But it does not follow that we should further restrict the information we do have in this specific way:<p>&gt; If they are not testing Chinese models with less than max thinking, then we should only compare them to other models also at max thinking.<p>If we have data only for “Chinese max” (let’s say it costs $1) and “non-Chinese low-to-max” (spanning $0.10 to $3), comparing them in full at least tells us what we can expect from each model if we have $1 to spend + the maximum performance we can expect with unlimited budget. What’s the point of restricting to “max vs max”, which answers only the second question?
            • paimapi35 minutes ago
              Usage? If most people predominantly default to Medium thinking level for usage then it would make sense to benchmark that first.<p>Another analogy for this are review sites overly focusing on reviewing all the SKUs of GPUs that cost $1k+ for an audience that, according to Steam&#x27;s Hardware Survey, predominantly use midrange X060s-equivalent cards. Sure, it&#x27;s fun to know whether or not you can enable PT on a 5090 but for most of us, covering 4070 TIs and below in-depth (meaning actually taking the card apart, seeing how good&#x2F;bad the past jobs are, possibly buying multiples of the same SKUs to evaluate QC, etc) is much more useful because of the value prop.
              • spider-mario8 minutes ago
                &gt; Usage? If most people predominantly default to Medium thinking level for usage then it would make sense to benchmark that first.<p>Sure, that’s not in question. The discussion here is not about what to measure (we agree that benchmarking only max is suboptimal) but how to compare models if you have an already fixed set of measurements, outside of your control, which includes only “max” for model A but a wider range from model B.<p>The person I was responding to was claiming (in their previous comment) that the sensible thing to do, then, was to artificially truncate the data for model B to “only max” as well.
          • tough2 hours ago
            sounds like context collapse
        • dtech8 hours ago
          This is a silly take. To continue the analogy vacuum A might need the power at 50% to vacuum something, B 100%. You cannot then test both at 100% and say &quot;both finished the task, but A required twice the resources so it&#x27;s worse&quot;, since you didn&#x27;t check what resources it <i>actually</i> needed.
          • throawayonthe8 hours ago
            isn&#x27;t that exactly the argument?
          • parineum8 hours ago
            The resources are dollars and the output is tasks. Choose whatever reasoning makes it work best and compare
            • PunchTornado7 hours ago
              Are the resources only dollars? We shouldn&#x27;t care about energy consumption? Water? It is difficult to calculate but we shouldn&#x27;t forget about it.
              • baq7 hours ago
                The idea is that dollars price in everything. It’s usually incorrect, but it’s also usually correct enough; and still much better than paying for services with water and joules directly
                • cyanydeez7 hours ago
                  Yes. The idea than money is equal to everything is kinda why we are all going to create the torment nexus. As it will be the most efficient way to torment civilization without all that effort.
                  • baq7 hours ago
                    Money helps allocate resources efficiently, that’s all there is to it. It’s about as good or bad as guns. If we didn’t have money, someone would invent it and in fact it was invented more than once in different cultures.
                    • cyanydeez22 minutes ago
                      I don&#x27;t deny that; but it rewires some peoples brains to the point that everything is adjustable via cash, including people, including their emotions, etc.<p>It&#x27;s very clear part of America&#x27;s diesease is the idea that everything can equate to a dollar cost.
              • parineum28 minutes ago
                Energy and water both cost dollars.
      • Aurornis11 hours ago
        The Artificial Analysis charts show max effort by default, which isn&#x27;t how most people use them.<p>Scroll down to the cost per task pareto front chart which includes more difficult levels by default: <a href="https:&#x2F;&#x2F;artificialanalysis.ai&#x2F;models&#x2F;qwen4-max#intelligence-per-task-tabs" rel="nofollow">https:&#x2F;&#x2F;artificialanalysis.ai&#x2F;models&#x2F;qwen4-max#intelligence-...</a><p>Qwen4-Max is not in a good spot on that chart. Astra xhigh and Fable 5.2 medium had a similar cost per task with higher intelligence ratings.
        • esperent10 hours ago
          I cannot see cost per task for any other thinking levels for Qwen 4 Max.<p>If it&#x27;s actually there, what is the cost at medium thinking? It would show Qwen 4 Max (medium) if it&#x27;s there.<p>Everywhere that I can see on the page it just shows ”Qwen 4 Max&quot; while for 6.1 Sol, Opus 5.5 there&#x27;s always a thinking level in brackets after: (low), (medium), (max) etc.
          • lambda9 hours ago
            They don&#x27;t always run the benchmarks at every reasoning level for every model; only a few prominent ones. They do it for the US frontier labs, plus a few of the Chinese labs, but not consistently. You can see that DeepSeek has a few reasoning levels tested, GLM 5.3 has a couple, as does Qwen 3.8 27B, but not other Qwen models, and a lot of the Chinese models they only run at max reasoning.<p>It&#x27;s possible that they will run Qwen 4 Max at other reasoning levels in the future, or it&#x27;s possible that they never will. Regardless, it&#x27;s unlikely to be better at any reasoning level on a price&#x2F;performance basis than the GPT 6.1 Sol&#x2F;GPT 6 Astra pareto frontier. It&#x27;s close enough to Claude Opus 5.5 line at lower reasoning levels that it could potentially cross that if it doesn&#x27;t loose too much performance at lower reasoning levels. But it&#x27;s far enough from the global pareto frontier that it&#x27;s unlikely that it would cross that at any resoning level.
            • esperent9 hours ago
              Right, so my original point stands: if they&#x27;re testing this model at only max thinking level, then it&#x27;s only valid to compare it to other models also at max thinking level.<p>You can see for example that the cost per task for 6.1 sol is about 4x higher at max compared to medium thinking level.
              • brainwad8 hours ago
                No, it&#x27;s only valid to compare to models at the same cost level, regardless if that&#x27;s their max level or not. Fable&#x2F;Astra having higher effort modes than Qwen doesn&#x27;t mean that you can&#x27;t compare Fable medium to Qwen max.<p>Which effort levels are cost-equivalent is debatable and depends on whether you are paying API prices or on a subsidised plan or are able to run the models yourself on your own hardware. But I think AA uses public API pricing for everyone as a sensible baseline.
                • esperent7 hours ago
                  If you look at the pareto chart for e.g. 6.1 sol you&#x27;ll see a curve of about 6 data points, some of which are on the Pareto frontier, some are not. For Qwen 4 max, many others, there&#x27;s just a single data point so we can&#x27;t just how close <i>any</i> thinking level would get to the Pareto frontier.<p>I&#x27;m not saying this model would land on the frontier - from the single data point it looks unlikely. However, that single data point might be making it look much further from the frontier than it should be.<p>EDIT: I will also take back my point that we should be comparing models at the same thinking level - max on one model won&#x27;t correspond to max on another. What we need is a range for each model. What we have is a range for the most popular models, showing that the same model with different thinking levels can have vastly different results. Then for less popular models we have just one data point, which isn&#x27;t enough to draw the kind of conclusions people are drawing in this thread.
              • aenis7 hours ago
                &gt; You can see for example that the cost per task for 6.1 sol is about 4x higher at max compared to medium thinking level.<p>It does not (universally) hold for long running workloads -- which is what people are increasingly doing.<p>Leave a Sonnet on medium, it will complete the task, spend more time and possibly significantly more token money, vs. Opus 5.5 on xhigh.
        • behnamoh9 hours ago
          &gt; The Artificial Analysis charts show max effort by default, which isn&#x27;t how most people use them.<p>I always use Max effort in both Opus and Fable. It&#x27;s actually faster overall because tasks get done faster with fewer &quot;corrections&quot; from my end.<p>On GPT models, I&#x27;d sometimes dabble in xhigh&#x2F;high but that&#x27;s because OpenAI usage quota are too limited.
          • esperent7 hours ago
            &gt; It&#x27;s actually faster overall because tasks get done faster with fewer &quot;corrections&quot; from my end.<p>Do you have proof of that or is it just vibes?
        • SV_BubbleTime10 hours ago
          How “safe” is Qwen4? Will it immediately shut down on security research like the US frontier models?
          • lambda9 hours ago
            Not sure about Qwen 4. I have a coworker who&#x27;s been doing some pretty good reverse engineering with Kimi K3.1 recently.
      • croon6 hours ago
        I assume those cost comparisons are on API token cost.<p>I&#x27;d love a comparison on for example monthly quota usage per task, between all ~$20&#x2F;month subscriptions.<p>It&#x27;s so difficult to compare how much &quot;productive&quot; usage you get out of each subscription, without signing up to each one at least.
      • aenis8 hours ago
        &gt; If they are not testing Chinese models with less than max thinking, then we should only compare them to other models also at max thinking.<p>Nah, there are interesting conclusions to be had in comparing various model&#x2F;effort level combinations, esp. when it comes to cost and time per task.<p>I have my own private benchmark suite; it involves algorithmic problems, the need for good visual understanding, and the ability to run long workloads unattended (as the agents need to run a few iterations, self-evaluate the results, and hand over the assignments when they feel they have succeeded). Chinese models do poorly on such problems, and running them at anything less than max efforts is pointless; they can&#x27;t even hand over the assignments graded against a mechanical set of tests they themselves can execute. Hand that same task over to, say, Sonnet medium and you have a weird outcome: excellent result, but way more time and tokens burned than, say, Opus 5.5 or Sol 6.1 running at high. I do those benchmarks just to be able to spot such patterns and then better choose the models for my real work - but i wouldn&#x27;t have caught it if I wasnt running various model&#x2F;effort combos.
      • lcnPylGDnU4H9OF11 hours ago
        &gt; If they are not testing Chinese models with less than max thinking, then we should only compare them to other models also at max thinking.<p>They compare different reasoning levels.<p><a href="https:&#x2F;&#x2F;artificialanalysis.ai&#x2F;models&#x2F;deepseek-v4-1-flash-1017" rel="nofollow">https:&#x2F;&#x2F;artificialanalysis.ai&#x2F;models&#x2F;deepseek-v4-1-flash-101...</a>
    • seizethecheese11 hours ago
      Does anyone really believe the cost data for benchmarks? Pretty sure all models will behave differently at this point when they get benchmark shaped questions.<p>(And I’m sitting on data showing Sol 6.1 and Qwen 3.8 Max cost roughly the same per task, for real tasks.)
      • eli11 hours ago
        The cost per task metric is accurate for those benchmark tasks. If your tasks don’t look like benchmark tasks..you should use your own benchmarks :)
        • XCSme11 hours ago
          Lol, I did that with <a href="https:&#x2F;&#x2F;aibenchy.com" rel="nofollow">https:&#x2F;&#x2F;aibenchy.com</a>
      • killingtime7411 hours ago
        I mean not believing intelligence score is reasonable but cost is just reporting what they spent?
        • seizethecheese11 hours ago
          I believe low cost but not high cost, since a model will work really hard when a question looks benchmark-like.
    • ctolsen6 hours ago
      vals.ai puts it at just about Astra level fwiw.
      • letrix16 minutes ago
        Also Sol is about Astra levels. And Sonnet is better than Opus.
    • scotty7910 hours ago
      Looks like a slightly weaker, cheaper Kimi.
    • spwa46 hours ago
      I wish they&#x27;d also post the benchmarks according to the previous &quot;version&quot; of artificial analysis.
    • augment_me10 hours ago
      Artificial Analysis is a VC-backed firm and it&#x27;s unclear what the incentives are there.<p>They have altered their benchmark multiple times in the last 4 weeks (v5 in <a href="https:&#x2F;&#x2F;artificialanalysis.ai&#x2F;articles&#x2F;artificial-analysis-intelligence-index-v5" rel="nofollow">https:&#x2F;&#x2F;artificialanalysis.ai&#x2F;articles&#x2F;artificial-analysis-i...</a>) and the last 5.0.1 change <a href="https:&#x2F;&#x2F;artificialanalysis.ai&#x2F;methodology&#x2F;intelligence-benchmarking" rel="nofollow">https:&#x2F;&#x2F;artificialanalysis.ai&#x2F;methodology&#x2F;intelligence-bench...</a> pushed up Anthropic&#x27;s models a bit, you can see that in v5.0.0 at the first link, Opus 5.5 has a lower score than in it gets in v5.0.1, and v5.0.1 was released just before Fable dropped.<p>Some part of me believes that the benchmark mix was put together to favor Anthropic&#x27;s models.
      • Grimburger2 hours ago
        &gt; Favor Anthropic<p>Rarely seems to make them look good whenever I drop in to the site. Far more expensive for little gain. Even within their own effort levels it looks bad, Fable xhigh is 50% more in cost than high for zero improvement across a range of tasks.<p>A tiny bump in intelligence score hardly seems worth it for cost, and honestly most people&#x27;s tasks are not cutting edge work.<p>Current benches on there seem to make Sol look the best, at a certain point the subscription subsidy tap will turn off and people will be in for a rough time paying full price.
      • DeathArrow7 hours ago
        If Artificial Analysis is not to be trusted, then which benchmarks are?
        • DCKing6 hours ago
          I don&#x27;t think there are any that can be trusted. There&#x27;s too much riding on these benchmark results so the labs find ways to game them either by benchmaxing or by influencing the format of the benchmark.
        • ctolsen6 hours ago
          Check others as well. <a href="https:&#x2F;&#x2F;www.vals.ai&#x2F;" rel="nofollow">https:&#x2F;&#x2F;www.vals.ai&#x2F;</a> has it just under Astra.
      • spwa45 hours ago
        I think you mean that benchmaxxing has turned into a business model, not for the Chinese models (maybe them too, of course), but for the AI labs, and artificial-analysis is not remotely the worst offender.<p>Right now the big labs are buying benchmark tasks from &quot;benchmark labs&quot; (or whatever they call themselves) for absurd amounts of money, and making sure they perform well on them.<p>The incentives this generates are <i>terrible</i>.<p>The big AI labs will only buy benchmark tasks that the frontier models perform poorly on (because they have trouble finding any public ones). So the tasks are being mass-written now directly based on the models they&#x27;re supposed to train. Obviously the math says you shouldn&#x27;t do this. If there is a causal connection between the benchmark and the model, you lose the normal guarantee that improving benchmark scores will improve model.<p>In fact improving such benchmarks may make the model worse. To understand that let&#x27;s just give a ridiculous example. I create tasks like &quot;1 3 5 11 13 17, give next number&quot;. The next number is 301. These are the house numbers on some random street in Brussels, Belgium (I&#x27;m guessing the big jump is because they &quot;merged&quot; 2 streets, and just added 300 to the numbers of the second part).<p>If the model trains on this, performance will actually decrease because obviously this is not how you <i>should</i> reason.<p>BUT making tasks under this &quot;impossible to reason towards&quot; part is of course how you make tasks that might even be reasonable, and of course frontier models are going to suck on tasks like this. So they sell!<p>BUT they suck! Finding houses is actually going to work a lot better if you ignore this fact until it becomes unavoidable in one out of 1000+ streets.<p>And yet this is what benchmark improvement has moved to. If you look at the terminal-bench 4.0 tasks you will see what I mean. Doing better on those tasks ... does not actually make a better model BUT you have to actually go read those tasks to realize that. A lot of the new tasks largely demand you make weird jumps in your reasoning not based on understanding the underlying theory. They rely just being able to predict what goes wrong in certain cases when theory would suggest it should work, and mostly it even <i>does</i> work but for some reason not acceptably. They reward not doing the full work, realizing what goes wrong and correcting, without understanding why the normal reasoning won&#x27;t work and correcting without knowing why. I find it very hard to believe this makes a model more intelligent. It rewards knowing the encyclopedia by hard, so to speak.<p>But it certainly makes it look better. Whereas the obvious, coding improvements, do not make models look better anymore.
      • puppymaster9 hours ago
        AA must be doing something right. Ever since they released the new benchmark, you see open weight China model fans complaining about it under every release thread. Probably because it’s hard to benchmaxxed against.<p>But admitting their model is weaker is a no go. So conspiracy theory here we go
        • dtech8 hours ago
          For what it&#x27;s worth, in our own metrics and benchmarks the intelligence is a pretty good match, but the cost per task is wildly out of whack. We see the open weight models like GLM-5.3-flash or Deepseek 4.1 Flash 1017 use a fraction of their tokens&#x2F;cost of the benchmarks, which changes the cost-benefit equation a lot.
        • rvba8 hours ago
          or maybe it&#x27;s the other way round, open weight models are under siege while closed weight models are defended
        • anon3738398 hours ago
          &gt; Probably because it’s hard to benchmaxxed against.<p>It&#x27;s mainly just a composite of existing benchmarks. It would be very easy to benchmax if, say, AA shared with a preferred lab which benchmarks it plans to upweight&#x2F;downweight in the next version. Just speaking hypothetically, of course.
          • puppymaster8 hours ago
            Hence the constant switching around and running the benchmark themselves. They have been pretty transparent about which benchmarks get added or removed with every release.
    • montroser12 hours ago
      Yeah, it&#x27;s in the same ballpark as Kimi 3.1. Both of which are obviously open weight, so that&#x27;s at least a big differentiator from Sol. Let&#x27;s hope against hope that the much smaller flash model might beat deepseek v4.1 flash and run for as cheap.
      • cyanydeez7 hours ago
        Eh, as long as they make qwen3.8-flash-next equivalent on less resources and faster, they&#x27;re winning.<p>It&#x27;s like arguing windows 11 just needs more bloatware.<p>These models don&#x27;t have to all in one every task in the world.
    • zkmon9 hours ago
      Why do we trust this? Do they open-source their test code?<p>Also, a generic benchmark might not translate well into what everyone needs. I keep building my own test cases that test models for my own work specifics. Only that benchmark is useful to me. I don&#x27;t need it to win a PhD math quiz.
    • jrflo11 hours ago
      It looks pretty bad on those benchmarks. Is it possible that this is due to the recent crackdown on distillation?
      • raincole10 hours ago
        I think it&#x27;s the most plausible explanation.
        • yanhangyhy9 hours ago
          i think they also has limited acccess to the latest GPU. It&#x27;s possible huawei chips is their only options now.
          • j2j89 hours ago
            They have plenty of access via data centres outside of China. The first woman Jensen was revealed to have sex with is the CEO of one such Singaporean provider to China. The US allows this.
    • thefourthchime10 hours ago
      [dead]
  • hauntingseaweed4 hours ago
    Anthropic deserves to go bankrupt for trying to ban open-weight&#x2F;open-source competition. I hope their IPO flops.
    • jari_mustonen2 hours ago
      Anthropic and its CEO are preaching about “safety”. Meanwhile, they created a model considered “too risky” and gave US government access to it. This is a government whose track record includes detonating nuclear weapons over cities full of civilians, deliberately exposing prisoners to sexually transmitted infections, and dumping weedkiller across Vietnam on such a scale that it caused over 100,000 birth defects.<p>The US government is the most reckless actor on the planet and Dario just couldn’t wait to give it a tool to create new diseases, new atomic weapons, new spying tools, new biological weapons, and whatnot.
      • andai1 hour ago
        &gt; The Army released S. marcescens in eight tests to determine the vulnerability to enemy attacks (U.S. Army 1977). In addition, the Army conducted field testing with B. globigii and Aspergillus. S. marcescens and Bacillus subtilis were also released in San Francisco and New York subway systems, respectively, to study their dispersion.<p><a href="https:&#x2F;&#x2F;www.ncbi.nlm.nih.gov&#x2F;books&#x2F;NBK233505&#x2F;pdf&#x2F;Bookshelf_NBK233505.pdf" rel="nofollow">https:&#x2F;&#x2F;www.ncbi.nlm.nih.gov&#x2F;books&#x2F;NBK233505&#x2F;pdf&#x2F;Bookshelf_N...</a><p>They also previously released bacteria from planes over their own citizens:<p><a href="https:&#x2F;&#x2F;en.wikipedia.org&#x2F;wiki&#x2F;Operation_Sea-Spray" rel="nofollow">https:&#x2F;&#x2F;en.wikipedia.org&#x2F;wiki&#x2F;Operation_Sea-Spray</a><p>See also: Similar &quot;experiments&quot; in the UK around the same time.
      • heap_perms2 hours ago
        Just stating facts. I don&#x27;t know why you&#x27;re getting downvoted.
        • mycall1 hour ago
          Biased, only shows bad examples without good examples the US government does.
          • TheSkyHasEyes1 hour ago
            The reply was in context of the post above theirs.
          • Dependance1 hour ago
            So those are okay to do, even to you, if watered down with enough &quot;good&quot; on top ? You sound like a partisan of the Leopard&#x27;s Eating People Face Party. &#x2F;s
        • WarmWash30 minutes ago
          &quot;I stopped wearing my seatbelt after reading about people who lived because they were thrown from their car.&quot;<p>A pile of facts don&#x27;t inherently build up to the truth. The sooner people learn this, the sooner we can leave behind so much bullshit and misdirection in society.
      • syndacks1 hour ago
        This is a government whose track record includes helping liberate Europe from the Nazis, rebuilding devastated countries through the Marshall Plan, putting human beings on the Moon, and funding scientific research that helped make the internet possible. Through PEPFAR, it has helped save millions of people from HIV&#x2F;AIDS.<p>The US government has been one of the most consequential forces for human progress in history. Giving it access to powerful AI could help accelerate medical research, strengthen public services, protect democratic allies, and make scientific breakthroughs that private companies would never pursue alone. Dario is right to recognize the potential: helping democratic government use extraordinary technology responsibly is an ambition worth having.
    • mattmaroon3 hours ago
      Don’t let the fact that they never did that and have explicitly said they’re opposed to it get in the way of your anger.
      • Iolaum3 hours ago
        They have been trying to do that. They never signed the open weights support letter and have been yapping about dangers and standards basically trying to shape the ecosystem on their view of how it should be; which would effectively ban open weight models.
      • isgb3 hours ago
        <a href="https:&#x2F;&#x2F;www.implicator.ai&#x2F;openai-anthropic-lobby-washington-open-weight-ai&#x2F;" rel="nofollow">https:&#x2F;&#x2F;www.implicator.ai&#x2F;openai-anthropic-lobby-washington-...</a><p><a href="https:&#x2F;&#x2F;thenewstack.io&#x2F;microsoft-nvidia-meta-and-open-weights&#x2F;" rel="nofollow">https:&#x2F;&#x2F;thenewstack.io&#x2F;microsoft-nvidia-meta-and-open-weight...</a><p>Then there&#x27;s all the &quot;open models are dangerous&quot; rhetoric while Anthropic and OpenAI&#x27;s models are displaying said danger. Then there&#x27;s the insane amount of money they&#x27;re spending on lobbying. On what exactly? Who knows, but since it&#x27;s their interest to kill open weight models and most companies tend to lobby to defend their own interests, it&#x27;s not an unreasonable assumption that some effort is going into that.
      • ddxv3 hours ago
        When did they say they oppose it? I&#x27;ve only seen them actively scare mongering that open weight models are too dangerous?
  • seizethecheese12 hours ago
    I hope they are able to serve 4 faster. Qwen 3.8 is a great model but I’ve regretted including it in the battle system for opencode [1] we have been building because it scores so low.<p>And it’s a shame because the only reason it scores so low is that it’s slooooow. (Like 30 tokens per second vs opus near 80.) If you exclude voting, Qwen 3.8 does well.<p>(We allow early voting contrary to Arena.ai because I think it makes sense for agents, where it doesn’t for chat. This means our data takes speed into account.)<p>Also, it would be ideal if they served this with third parties. Part of the slowness with 3.8 is just not a lot of providers. Their open source version is much faster, but we couldn’t include it because it doesn’t handle images (another own goal by Alibaba).<p>[1]: <a href="https:&#x2F;&#x2F;agentduel.com&#x2F;leaderboard" rel="nofollow">https:&#x2F;&#x2F;agentduel.com&#x2F;leaderboard</a>
    • jakswa11 hours ago
      In case it translates to Qwen 4 27B, I&#x27;ve been running Qwen 3.8 Flash Next (which while MoE (and bigger) is supposed to be a preview of Qwen 4&#x27;s architecture) and it runs pretty solidly regardless of context window size and completes random deepswe challenges I throw at it locally (at Q2-Q3 quantization! so it fits on my shitty setup).
      • wronglebowski2 hours ago
        This, I continue to be amazed using Strata to serve IQ3_XXS. It&#x27;s such a capable model for something that runs on (relatively) little hardware.
      • arlcode5 hours ago
        That is our experience as well at FP8 for 3.8 FN.<p>For many of our use cases it&#x27;s currently far into the &quot;good enough&quot; range because external factors (such as process design) are now blockers.
    • FrostKiwi11 hours ago
      &gt; but we couldn’t include it because it doesn’t handle images<p>Maybe I&#x27;m mixing up model variants, but Qwen 3.8 27B that I run locally totally does handle images. Surely the bigger one would as well?
      • seizethecheese11 hours ago
        The model I’m referring to that would be faster and similar size, but doesn’t handle images is Qwen 3.8 2.4T: <a href="https:&#x2F;&#x2F;openrouter.ai&#x2F;qwen&#x2F;qwen3.8-2.4t-a95b" rel="nofollow">https:&#x2F;&#x2F;openrouter.ai&#x2F;qwen&#x2F;qwen3.8-2.4t-a95b</a>. It’s weird that this is pretty much the same as Max but doesn’t support images. No idea why.
    • forgingahead12 hours ago
      Qwen3.8 doesn&#x27;t handle images?
      • seizethecheese12 hours ago
        Sorry that was confusing. Qwen 3.8 2.4T is faster than what we have on leaderboard but it doesn’t handle image input: <a href="https:&#x2F;&#x2F;openrouter.ai&#x2F;qwen&#x2F;qwen3.8-2.4t-a95b" rel="nofollow">https:&#x2F;&#x2F;openrouter.ai&#x2F;qwen&#x2F;qwen3.8-2.4t-a95b</a>
    • zdragnar8 hours ago
      Is the speed due to a difference in the way it &quot;reasons&quot;? I was trying out a local version (3.8 27b) and watching the thinking, it&#x27;s absurd how often it challenges itself. Simple facts get questioned multiple times, and it has to keep reminding itself that it already established some things as facts earlier.<p>Instructing it to ask any and all questions as necessary sends it into an existential crisis, because the question tool (oh-my-pi harness) suggested limiting the number of questions, and it couldn&#x27;t figure out how to do both.
      • cyanydeez6 hours ago
        Aside from making our engineering prompt terse, I place a token budget on the thinking traces. Those traces are mostly meaningless(in the reasoning realm, not the usefulness realm), and just tell it to do something productive. For coding, this just makes it read doc, tests, etc.<p>I believe that&#x27;s all reasoning levels do, but most inference engines should have thinking budget.
    • motahir_alam6 hours ago
      [dead]
  • pu_pe7 hours ago
    I think Alibaba is not aiming for the frontier on this class. Their previous Qwen3-Max was released in September, right around when Opus 5.5 was coming out, and it was nowhere close to it as well. However, one or two tiers below (Flash, 27B) they really shine and are SOTA or close to it.<p>DeepSeek has a similar problem, I feel like their full model is hardly even worth using, while the Flash tier is absolutely amazing.
    • thibsrn2 hours ago
      It seems to be the case. I would qualify them more as flagship-killer than frontier, using the same terminology used for smartphone classification. As opposed to the big American shops which focuses solely on performance, they have operational costs and other factors in their constraints and requirements. As soon as the market reaches some maturity, this will prove to be very important.<p>Crisis sold far fewer copies than Call of Duty.
    • mrklol7 hours ago
      I am unsure. Why shouldn’t they? Why even work on a big model (especially training time) if “nobody will use it as competitors are better”? We’re still talking about nearly 2.5T params here. Deepseek’s pro model is still heavily smaller and like you said isn’t directly worth using, obviously it has the open source &#x2F; self hosting advantage.<p>Imo there’s a need and they try to push here, it’s also giving them advantages for their smaller models but if the big models never perform they will have a bad time or maybe should even stop.
      • namr20002 hours ago
        If you&#x27;re aiming to make the best possible small model, you still need to train a larger model and then do logit-level distillation from it. This is where most of the gains in small models have come from.
      • cyanydeez6 hours ago
        My use of qwen3.8-flash-next suggests they&#x27;re trying to make a low cost model that knows how to program and they think the &quot;total addressable market&quot; is serving more people at lowest inference costs possible.<p>Think of it like a hind brain+fore brain discussion. Instead of focusing on making some ultra model that can do everything but costs a ton, you have your smaller models gather the bits and feed it better data<p>Agentic coding is the target then. If they want their work force to not care about token gen, that&#x27;s what you do.<p>As an analog, they&#x27;re making a drone swarm while US is making 5th Gen fighters. Completely different battle fields.
        • pu_pe4 hours ago
          Yeah, I feel like models like Qwen3.8 27B are making clear trade-offs, like having very little world knowledge baked in. I believe those are intentional, so they can focus more on other areas. I guess that penalizes them on these benchmark aggregators.
        • anentropic5 hours ago
          The &#x27;frontier&#x27; incorporates two dimensions though, usually score x cost, so it&#x27;s not just the best score.<p>Meaning models below the frontier either cost more for same score or score worse for same cost.
          • cyanydeez24 minutes ago
            I think the argument is like MacOS vs windows vs linux. AI models from anthropic&#x2F;openAI are windows and MacOS; they&#x27;re plat forms for &quot;skilless&quot; to start on and of course you can gravitate towards the development end. They&#x27;re bloated in that sense.<p>Open models follow the linux goals which is small tools that chain together. If I can get a small model to understand how to program for another model or do the programming bits for a large model, then the trade off is flexibility. You can imagine your giant AI that can do anything, like windows&#x2F;Mac is to your benefit, but there&#x27;s a clear flexibility line to how these models act.<p>Even if it takes longer for my local model to find something, that doesn&#x27;t mean it&#x27;s taking a long time because it&#x27;s doing wrong things.<p>Honestly, I think a lot of people using SOTA don&#x27;t understand what you can do with a proper engineering harness; we all live in a yellow submarine.
  • dan_thinks13 hours ago
    This is exciting. The continued focus on efficiency alongside quality is encouraging - it makes me hopeful to see what smaller models will become capable of.<p>Looking forward to the &quot;Flash&quot; and 27B versions.
    • neurophant1 minute ago
      Waiting for the Flash tier too. My experience running coding agents on cheap vs big models: the cheap one gets the same work done, roughly twice as slow. The big model buys you speed, not results.
    • teaearlgraycold10 hours ago
      I&#x27;m also hoping someone puts out a competitive 30-70B model soon. There&#x27;s this massive dead zone between 24GB and &lt;100GB of VRAM where it seems your best option is Qwen 3.8 27B.
      • dstu4 hours ago
        The 3.8 Next-Flash release is comfortably the best model on a 96GB Mac Studio. I’ve tried various engines and Sushi (<a href="https:&#x2F;&#x2F;github.com&#x2F;beamivalice&#x2F;sushi" rel="nofollow">https:&#x2F;&#x2F;github.com&#x2F;beamivalice&#x2F;sushi</a>) is the most performant so far.<p>Hopefully the Qwen 4 Flash model will still fit within 96GB without requiring excessive quantisation as, verbose thinking aside, 3.8 is a very solid workhorse model.
        • samtheprogram48 minutes ago
          Actually, that’s a good thing. You can do multiple requests at once. Sub-agents. Etc.
      • shockembopper10 hours ago
        There are many engines that alleviate this issue in particular. The current mood is running qwen flash next using strata, basalt, ect.
      • q176 hours ago
        Same, I recently found Nemotron 3.5 74B-A9B really good at filling this gap though, could do (very slow) local inference on just 32GB DDR4, without the quality dropping to the floor
  • andy_ppp3 hours ago
    Quite a few saying that they tried earlier snapshots of this model and are finding it does not live up to expectations. Firstly do other agree and also why would the model get worse - could they be doing something weird to the model that is making less aligned at the end of training, potentially for political reasons?
  • akulbe42 minutes ago
    I subscribed to the Lite plan, but I&#x27;m not finding where I can get&#x2F;generate an API key, to be able to use it with something like Hermes Agent?
  • nelaggy12 hours ago
    is the open weights lag widening? these open models don&#x27;t seem to be improving as quickly as ant&#x2F;oai
    • julianlam2 minutes ago
      I think looking at the lag between frontier and open-weight is looking at an uphill battle and assuming it&#x27;s equal footing.<p>Lots of research goes into open-weights, optimizing cache re-use, storage, inference costs, etc., and all these benefits are easily rolled into closed AI models. Meanwhile any advances made by OpenAI, Anthropic, etc. are kept hidden away to give their models an edge.<p>It&#x27;d already be a miracle if open-weights were gaining on-par with closed-weight models. That they&#x27;ve been closing the gap for most of the past year is wild!
    • rbehrends10 hours ago
      No idea, but this is a rapidly changing field. Opus 4.5 was released a little less than a year ago and it feels shockingly limited compared to even current open flash models. (I literally just tested it the other day.)<p>So, with new models coming out on a weekly basis, I think trying to figure out a future trajectory is just going to be a wild guess. For all we know, Anthropic and OpenAI models may also be plateauing in the near future; or there may be a breakthrough and they may improve beyond our wildest expectations. I honestly don&#x27;t know.<p>That said, I suspect available compute and model sizes that major Western vs. Chinese labs can handle will still keep the gap open at least in the short term.<p>On the other hand, I am also not sure how much the gap will matter past a certain point. I am specifically wondering at what point people will become interested in smaller models fine-tuned for specific application areas (e.g. cheminformatics) rather than massive generalist models.
    • x31211 hours ago
      I can&#x27;t imagine them keeping up long term given the information asymmetry plus compute restrictions on top of it.<p>I feel open-weight was able to keep up the past year using efficiency improvements (hybrid attention, improved MTP drafters, improved optimizers) to offset having less compute. But OpenAI&#x2F;Anthropic started cutting token costs just a few months after these, which to me suggests they copied these improvements. Indeed it would be hard to work in LLM training and not know about DSpark, Muon etc.<p>So what do you do if closed source uses your learnings, gives no knowledge back, and has 10x more compute and funding on top of it? The situation feels quite dim right now, esp. with the Nvidia thing.
      • gimlichgu11 hours ago
        open always wins
        • pell11 hours ago
          Right now open is heavily subsidized by China and I fail to see what would happen if that stopped.
          • koe1238 hours ago
            You are subsidizing American AI with your pension after the IPO
            • pell3 hours ago
              Obviously American AI is similarly geo politically supported. However the business model so far is at least somewhat clearer. They invest in the cutting edge and want to sell access to it.
              • jcmfernandes2 hours ago
                American AI has a clear business model: selling you inference.<p>Chinese AI also has a clear business mode: making money from its (i.e., chinese AI) applications.<p>I&#x27;ll argue that the latter is closer to the end-goal here than the former.
        • jrflo10 hours ago
          What advantage do they have over closed? You may want open to win, but logically I don’t see how it would
          • EddieRingle9 hours ago
            I can run an open model on my own hardware, in my own home (which has solar panels on the roof, so electricity usage is mostly a non-issue). Or, if needed, I can rent compute from somewhere like you&#x27;d do for any other workload.<p>Once I have a model locally, I know it isn&#x27;t going to change on me. Compare that to using an API from Anthropic or OpenAI (and yes, the only thing you can say is you&#x27;re using an API, not one model), they can discontinue a model you&#x27;re relying on (as every model could have peculiarities you&#x27;ve designed around) or modulate their speed&#x2F;quality day-to-day to save on compute costs.<p>And trying to claim that buying your own hardware doesn&#x27;t pay off compared to API or subscription costs doesn&#x27;t hold much water either, at least not for me. For starters, I already had the GPUs and machines I started running models with, whether it was for gaming or for other reasons. And in the case of the new GPUs I bought recently (2x AMD Radeon Pro R9700; no you do not need CUDA and you haven&#x27;t for a long time now), I can hold a reasonable expectation that they will hold at least some of their value years from now, compared to paying for API&#x2F;subscription usage for years in which case you&#x27;re left with nothing.<p>So, if you have a local model that is &quot;good enough&quot; (and they already exist for many use cases), it&#x27;s just kind of the obvious choice to make, IMO. And in just one person doing it as a hobby while I&#x27;m still looking for a job 16+ months after a layoff. A small business and certainly a large enterprise has even less reason to open themselves up to all the downsides that come with using Anthropic or OpenAI.
          • hellcow10 hours ago
            Local models already crossed the threshold where they&#x27;re quite good and useful for many things, even with cheap setups. I walked away really impressed with (non-coding) use of Byteshape&#x27;s Qwen3.5-9B quant on my 8GB RTX 3070 for instance. The model was fast, helpful, and 100% private.<p>If we extrapolate over the next several years, prices for VRAM&#x2F;RAM will start to come down as manufacturing capacity comes online, models will continue to become more token efficient, and intelligence will improve at every quant size. As this happens, more and more people will be running LLMs locally on hardware they already own, and there&#x27;s going to be a point where local models can do everything you need to do unless you&#x27;re literally trying to do breakthrough mathematics research.
            • forrestthewoods9 hours ago
              I don&#x27;t think &quot;useful for many things&quot; qualifies as &quot;winning&quot;. It&#x27;s about two orders of magnitude shy of winning.<p>&gt; and there&#x27;s going to be a point where local models can do everything you need to do unless you&#x27;re literally trying to do breakthrough mathematics research<p>Maybe. But competitive markets pay exponentially more for incrementally superior services. Professional athletes are a great example of this. If you&#x27;re building a product with competitors and those competitors have frontier models and you stick to local then you&#x27;re probably going to lose both today and in the near-ish future.<p>If you&#x27;re bitter lesson pilled then the alpha is entirely in having a better model, more data, and more compute. Your vision is that human ingenuity will provide more alpha. Oof. I wish. But my tastebuds are very very bitter.
              • byzantinegene9 hours ago
                &gt; If you&#x27;re building a product with competitors and those competitors have frontier models and you stick to local then you&#x27;re probably going to lose both today and in the near-ish future.<p>i highly doubt this premise. the success of a product has very little to do with the amount of the intelligence used in producing it. you could throw millions of dollars using AI to develop a movie and it would still flop in comparison to an indie-made film.
                • wolfinterval3 hours ago
                  I think the premise is confusing what the amount even is in this context.<p>I want to use &quot;more intelligence&quot; but not in terms of single higher IQ.<p>I want to use more intelligence in terms of a sum quantity of knowledge work being performed. A team of 10 instead of a team of 5.<p>This is when a local model will shine. A fixed cost and I don&#x27;t have to worry about how much I am using it. I already can&#x27;t tell any difference at all between Opus and Fable. It won&#x27;t be that long that I won&#x27;t be able to tell the difference between a local model and Opus.<p>There will be an infinite demand for &quot;more&quot; intelligence but I think many people are making a category error on the dimension in question.
              • gpt59 hours ago
                Also - the “open” here doesn’t have the advantage of open source, where it can be improved by the community. At best, the community can tweak it a little. For actual improvements you need to wait for the next model which is trained by companies for billion of dollars, and could stop releasing them at any moment.
                • pyrale8 minutes ago
                  It can, through better quantizations and reinforcement learning.<p>Just this week, qwen 3.8-27b got two new interesting quantizations, one of which came from community. Llama.cpp also improves continuously, and novel alternatives like Strata pop up.
              • avadodin6 hours ago
                you don’t even understand the lesson, oofu.<p>the lesson is that training a model on all input ever output will never be as good as training a model on arbitrary outputtable input.
          • indy4 hours ago
            They help to prevent the &quot;we&#x27;ve always been at war with Eurasia&quot; problem that would occur if we depended on a couple of &#x27;approved&#x27; models
      • alfiedotwtf9 hours ago
        It all depends on if continual intelligence with each version is logarithmic or exponential..<p>.. if exponential, they’ll never catch up but if logarithmic, closed models will plateau and so catching up is inevitable.<p>My money is on a plateau
      • koe12311 hours ago
        Show us your stock holdings
    • Aurornis10 hours ago
      This is Qwen4-Max, which is not open weights.<p>Their open weights models are the the smaller ones, which aren&#x27;t released yet.
    • baoooooooooooo11 hours ago
      I think so but it may be a very short term effect, I think their pace has roughly matched the pre-speed up pace of the frontier labs (the 1.7-2x speed up noted on by METR), and I’d guess we see a similar speed up from the Chinese labs once their internal models hit an equivalent level enabling the same speed up.<p>Following that logic the gap widens a little bit here but then will steady again. Before this there has been a general catch-up trend too which probably also persists leaving us in a slow closing of the gap as the net effect?
      • gpt511 hours ago
        If the speed up is the reason, then they will never catch up (with traditional means). Because it means RSI predictions are real, and we will see an explosive growth in capabilities (far beyond exponential).
        • orangeboats8 hours ago
          &gt;If the speed up is the reason, then they will never catch up<p>This doesn&#x27;t take other factors into account. Compute capability, infrastructure capacity (the chips can&#x27;t work without electricity!) etc.<p>Those factors translate into a ceiling for this hypothetical exponential speedup.
          • gpt58 hours ago
            If compute is the bottleneck we’ll encounter in the coming year, then you would expect the gap to increase even more. The US and its companies has three times as much AI compute as the rest of the world combined.
            • orangeboats7 hours ago
              That is if you expect China and US to maintain the same compute gap over time. I wouldn&#x27;t be so optimistic if I were you.
              • gpt54 hours ago
                I’m actually expecting it to expand. There is so much new production being built right now by every company due to the shortage, and China doesn’t have a single EUV production, and only expect a prototype one in two years.
    • flincode2 hours ago
      I feel like they are improving what actually matters for LLMs, lower memory usage, compute. While we don&#x27;t know how much far ahead closed source models are in this, but it could be similar they might lower the price to subsidize or make it seem like its all good, either way tech like these being open means it will help us reach further in the long term.
    • taneq12 hours ago
      Might just be the commercial offerings scaling up parameters so much? Like for like the open models don’t seem to be lagging.
      • nelaggy11 hours ago
        either way, ant&#x2F;oai models dominating the intelligence-cost frontier is not a great outcome for those of us who prefer open weights
        • omnimus7 hours ago
          Even smaller open weights models are already great enough for much of &quot;assist human&quot; work.<p>This forces ant&#x2F;oai to lower their prices but how much that is subsidized nobody knows. It might very well just be they are in deep trouble, they see their clients leaving, they have to lower their prices.<p>But let&#x27;s not forget openai is riding on the edge of bankrupcy, their last investment round apparently gave them few months of runway, they are leaking money. I don&#x27;t think they know how to turn this around with all the open models being competitive.<p>I would just enjoy cheaper model costs while they are like this.
    • manmal10 hours ago
      Ant and OAI hiding their reasoning traces probably doesn’t help.
    • raincole11 hours ago
      Yeah, it turns out that the distillation theory has legs.
    • gimlichgu11 hours ago
      ant&#x2F;oai have stalled in the past 6 mos,
      • manmal10 hours ago
        Do you use them? Opus 5.5 is 2 months old and was a big leap forward. Sol is 4 months old, and was at least a 30% improvement.
  • WASDx4 hours ago
    Warning: The website downloads more than a gigabyte of videos before you even start watching anything.
    • mt_2 hours ago
      Alibaba benchmarking their CDN in prod.
  • husseinhb4 hours ago
    I was waiting GLM from z.ai and it still not releasing, almost all the models releases their new version of models and they still on 5.3 who is very weak in complex tasks and do half of the work and forgets to make migrations and like that issues. They sell us the yearly plan and stopped improving. Because of that i am just using in coding opus 5.5 and in chat fable they are the best models from experience.
  • xpct7 hours ago
    I was waiting for whether there&#x27;ll be Chinese model drops before the Anthropic IPO. Still a few days left, wonder if there&#x27;s more coming.
    • Havoc1 hour ago
      GLM will drop something soon I reckon
  • jhonof11 hours ago
    Any info on when the 27B or FN Qwen 4 models are coming out?
    • baoooooooooooo11 hours ago
      The past few days I’ve been finding myself asking “seriously, where <i>is</i> it”… feeling that the world has somehow changed because Qwen 4 27B was later than I expected.<p>Hoping it’s a great local model though!
      • kinleyd7 hours ago
        I feel quite the same way. I&#x27;m feeling better now since Qwen4 Max is out - 27B should be around soon.
    • gardnr9 hours ago
      Soon?<p><a href="https:&#x2F;&#x2F;github.com&#x2F;QwenLM&#x2F;qwen-code&#x2F;pull&#x2F;15762&#x2F;changes#diff-340d15cc12168e03631d00e9f5e3ec106f8db6f20657ebc1e187687d50ac5a1cR270" rel="nofollow">https:&#x2F;&#x2F;github.com&#x2F;QwenLM&#x2F;qwen-code&#x2F;pull&#x2F;15762&#x2F;changes#diff-...</a>
      • nout8 hours ago
        The openai reference there is interesting
        • HenryMulligan36 minutes ago
          Doesn&#x27;t that just refer to using the OpenAI response&#x2F;completion&#x2F;etc. API? Just about everybody (llama.cpp, vLLM, etc.) supports that. I think the reference to &quot;qwen4.5-max&quot; is more interesting myself.
    • NorwegianDude6 hours ago
      It was delayed, so who knows. Alibaba planned on at least showing it off last week, but they changed it.
  • anana_12 hours ago
    Artificial Analysis score doesn&#x27;t look too hot. Doesn&#x27;t even surpass Kimi 3.1
    • rbehrends12 hours ago
      While I doubt that Qwen 4 Max can seriously compete with frontier models, the AA index has become all but useless, due to the introduction of new benchmarks in the aggregate which are of questionable value as a measure of &quot;intelligence&quot; [1]. AA has seemingly also abolished updating the Coding Agents Index, which was an okay benchmark for coding ability, and older models are completely absent from the v5 version of the full index.<p>[1] Especially Terminal-Bench-Science 0.1 stands out. Example problems:<p>- <a href="https:&#x2F;&#x2F;hub.harborframework.com&#x2F;tasks&#x2F;terminal-bench-science&#x2F;rdkit-ic-constraints" rel="nofollow">https:&#x2F;&#x2F;hub.harborframework.com&#x2F;tasks&#x2F;terminal-bench-science...</a><p>- <a href="https:&#x2F;&#x2F;hub.harborframework.com&#x2F;tasks&#x2F;terminal-bench-science&#x2F;cilia-segmentation" rel="nofollow">https:&#x2F;&#x2F;hub.harborframework.com&#x2F;tasks&#x2F;terminal-bench-science...</a><p>Probably very relevant to people working in these fields, but it is honestly not surprising if a lab with limited compute will not train for these technologies, and it says pretty much nothing about more common use cases. I also suspect that this benchmark measures knowledge retrieval more than raw reasoning ability. This is where you really need a breakdown of capabilities and not an increasingly meaningless aggregate number.
      • gardnr10 hours ago
        Did AA take on funding or something?
        • alfiedotwtf9 hours ago
          Not only that. I find it weird that some models like Laguna S 2.1 were never included.<p>Maybe somebody should archive the scores and see if they in the future ever backtrack particular models to favour others (like last week)
      • izabera11 hours ago
        alibaba has &quot;limited compute&quot;?
        • rbehrends11 hours ago
          Compared to OpenAI and Anthropic, yes. There&#x27;s this export ban on Nvidia chips, after all. And all the CPUs they have won&#x27;t help much with model training.
          • mitthrowaway29 hours ago
            Can&#x27;t they rent GPU time in American data centers like everyone else does?
            • j2j89 hours ago
              They do in huge Singaporean ones like Megaspeed.
        • mrheosuper5 hours ago
          That and limited training data.
          • throwa3562622 hours ago
            I am surprised that this is a problem for China.<p>CCP can probably &quot;volunteer&quot; a couple of thousand to build a high quality national training data bank.
    • anon37383912 hours ago
      AA has become pretty meaningless. They just keep adjusting the recipe to get a preferred ranking result, with no supporting evidence for their decisions.<p>Edit: In addition to reshuffling the weighting of individual benchmarks, they also do things like test Anthropic and OpenAI in their native harnesses, while not doing the same for Qwen&#x2F;Deepseek&#x2F;etc.
      • stymaar8 hours ago
        &gt; AA has become pretty meaningless<p>AA has always been meaningless, but people keep referring to it anyway just because the synthetic view it provides is too convenient (even if meaningless).
      • gr_norm12 hours ago
        Yeah, I don&#x27;t know why people bother with it anymore. It no longer reflects what you&#x27;re likely to see in real-world usage, at least IME.
      • the_duke12 hours ago
        To be fair, a lot of the popular benchmarks have become pretty meaningless.<p>Models often get great scores despite being a lot worse than the frontier in practice.
        • stymaar8 hours ago
          A benchmark is only relevant if it was released <i>after</i> the model it evaluates.
        • anon37383910 hours ago
          This is true, but if you look at the change in weighting to specific benchmarks from one version of AA to the next, it looks <i>really</i> arbitrary.
      • wasabi9910119 hours ago
        I thought they were an independent company, what motive would they have to prefer some ranking results over others?<p>Edit: I do agree their harness methodology is pretty bad, but I think that&#x27;s more directly attributable to the fact that harnesses appeared after their initial methodological design, and they are too lazy or resource constrained to improve their methods. Still dishonest, but a simpler motive.
      • stratts12 hours ago
        One thing I find confusing about AA is that they show GPT 6.1 dominating the lower end of the Pareto line, with eg, GPT 6.1 Sol at medium supposedly costing far less per task than DeepSeek V4.1 Flash at high.<p>Yet every time I compare the two in something like Pi, Sol ends up being 5-10x the price for a simple query, something like 7-10c vs. less than 0.5c for DeepSeek. What&#x27;s the deal?
        • blackoil12 hours ago
          Same for me coding tasks taking $20-50 are easily done by deepseek in $5-10
    • selectodude12 hours ago
      Anthropic finally figured out how to block them from distilling Opus.
      • m00dy12 hours ago
        how so ?
        • gardnr10 hours ago
          Anthropic has been trying many different ways to prevent this, including a proposal to nerf your sessions if they “detected” malicious activities (anyone’s coding sessions would “degrade” if detected as malicious). This led to a huge outcry from developers complaining that they “need to be able to trust” the responses from Claude and intentionally poisoning responses in some sort of weird shadow ban was unacceptable. They walked it back but the codemap leak also showed some dubious weird regexes to nerf “malicious” users. They have a real distillation problem. It’s hard to have empathy for a trillion dollar company that distilled the contents of the internet into a thing and now is crying foul when someone distills their outputs.<p>There was also a bug in which the reasoning tokens were “encrypted” for opus but you could ask Haiku to regurgitate them and it would.<p>Anthropic have been doing anything and everything to prevent distillation “usage” and it’s caused me at least once to cancel my subscription because it’s hard to trust a company that secretly alters its responses based on something I can’t see and never get feedback on.
          • m00dy9 hours ago
            what do you think the current status of this like encrypting reasoning tokens along the way ?
        • BoorishBears5 hours ago
          This weekend I had the audacity to use Claude Code to help with the code side of post-training a model...<p>Asking it to think&#x2F;reason too strongly would hit classifiers. Quoting reasoning summaries in my messages would hit classifiers. It thinking about how post-training effects another model&#x27;s would hit classifiers.<p>I thought it might be overzealous Fable 5.2 protection, but at one point I went through a chain of Fable -&gt; Opus -&gt; Sonnet. Eventually even Sonnet triggered a classifier no matter how the conversation went... I had to rewind and have Claude prompt Codex going forward for most of the work.
      • omnimus7 hours ago
        Of course all the work has always been done by US companies. China labs are just copycats.
    • AAisthetiobeofA12 hours ago
      &gt; Artificial Analysis<p>Ah yes, the reverse-benchmaxers of OpenAI models.
      • yipinwong11 hours ago
        Ah yes, another new account without any backup to the claim.
  • drnick18 hours ago
    Are these (hosted) Chinese models still accepting arbitrary cybersecurity and reverse engineering tasks? I am increasingly frustrated with Claude&#x27;s safeguards (and Codex, to a much lesser degree). It is flat out refusing to look at game binaries and anti-cheats for the sake of learning and devising Linux compatibility workarounds. It seems like in the near future people who want to do this type of work will need to buy a couple of DGX Sparks or retired datacenter GPUs and run abliterated open models on their own hardware.
    • Hackbraten42 minutes ago
      Anecdotal but qwen3.8-max has been grinding away at any RE task I’ve thrown at it since I switched from Claude three months ago.<p>(May not be representative because I’m extra cautious to frame all my RE requests explicitly as interop issues. Which they are, more or less.)
    • ctolsen6 hours ago
      Hosted where? In my experience they might put more guards when hosted by the company that makes them but if you use them through inference providers they do whatever you want.<p>Although, can’t say I’ve tried too hard (not keen on sending too much data to China) but I’ve been using K3.1 through Moonshot and I have had zero issues.<p>Mistral Large 4 (through Mistral) also doesn’t refuse a thing.
    • Havoc1 hour ago
      Latest kimi had cyber nerfed intentionally<p>Not blocked just less training
    • nolok8 hours ago
      If they don&#x27;t, they all have third party made abliterated versions available
  • Luke26426 hours ago
    I wonder if they could benchmax Program Bench, it&#x27;s one of the few benchmarks that looks almost exactly like a good fine tuning curriculum which might help general coding ability and agential work across the board?<p>I suppose it could make actually worse at taking instructions and redirection mid-task, but that could be easy to work around too. Just hold back a bit of the spec and drip feed it in later?<p>I don&#x27;t mean cheat on it, I just mean optimize for it harder!
  • m00dy12 hours ago
    I bet most of us waiting for 27b model and want to see how good it will be.
    • jtbaker11 hours ago
      I’m a QFN dude myself, but yes can’t wait to get it loaded on a spark.
    • kylehotchkiss12 hours ago
      M5U locked and loaded.<p>I wish there were more ~60b dense options like ripple too!
      • asimovDev7 hours ago
        I really want a M5U but man that price is hard to justify for a base model when I paid less than half that for my 96GB macbook.<p>Do you run Flash Next or 27b on yours?
    • McGlockenshire12 hours ago
      My 3060 yearns for the next generation MoE!
    • deebuildthings11 hours ago
      [flagged]
    • q176 hours ago
      Forget about 27b when are we getting new 4b, 9b and moe qwens, i&#x27;m dying out here
      • pferdone4 hours ago
        what about Bolt? so far I enjoy it alot on my NAS
    • neutrinobro11 hours ago
      Spooling up the V100s right now...
  • spwa46 hours ago
    So what is happening is that models are broadening difficult-to-reason-to skills on benchmarks. If you take the traditional benchmarks (ie. DeepSWE, Terminal-Bench slightly older version, GDPVal, Automation Bench, ...) this model is between Sonnet and Opus 5.5, VERY close to the frontier.<p>What has happened to artificial analysis is that they&#x27;ve begin to focus a lot on very broad tasks like legal reasoning, using very difficult to work with scientific software packages, correctly analyzing biology experiments (without or only tiny amounts of explanation, count cells, BUT leave it to the model to figure that cell that aren&#x27;t round, or have a distorted cell membrane (ie. have burst) are dead), very difficult physics reasoning (&quot;2 electrical wires are close together with power flowing in the same direction. What happens?&quot;, and you get points deducted if you don&#x27;t realize the problem with high voltage AC power, where the wires will shake themselves loose if the phase doesn&#x27;t match perfectly, meaning the wires need to be EXACTLY the same length or ...). If I&#x27;m being impolite I&#x27;d say the frontier models are trying to make sure they&#x27;ll pass the &quot;gotcha&quot; interview questions for many jobs.<p>The 3 differences:<p>Qwen 4 is a huge jump from Qwen 3.8 in coding<p>Qwen 4 is a big jump in using it for agentic workflows<p>Qwen 4 also performs much better on avoiding hallucination<p>If the difference between Qwen 4 Max and Qwen 4 flash is similar to 3.8 and 3.8 flash next (ie. about 10%), then 3.8 flash next may just barely beat Opus 5.4 at coding.
    • zozbot2346 hours ago
      &gt; So what is happening is that models are broadening difficult-to-reason-to skills on benchmarks. ... If I&#x27;m being impolite I&#x27;d say the frontier models are trying to make sure they&#x27;ll pass the &quot;gotcha&quot; interview questions for many jobs.<p>These &quot;gotcha&quot; skills will be needed for frontier research&#x2F;scholarly tasks. It makes sense to focus on these now that traditional domains like coding or frontend design are broadly saturated (i.e. Astra- or Fable-level outcomes) even by lower-tier open models.
      • spwa45 hours ago
        Real &quot;gotcha&quot; skills, yes (perhaps, they&#x27;re gotchas because the underlying theory is not well understood. More can be achieved, certainly for superintelligence, by forcing understanding of the underlying theory rather than the gotcha). AND the incentives are totally misaligned.<p>&quot;Gotcha skills&quot;, if we&#x27;re going to call it that, are so popular because the AI labs are buying &quot;tasks&quot; to train on at huge prices, but mostly if they&#x27;re gotchas.<p>The problem with that is that a collection of facts in the positivist sciences (and math) are always incomplete and almost preclude a full understanding of math, because it takes away the reason (to an extent) for full understanding.
  • libpcap11 hours ago
    Does it provide an objective response when asked, &quot;What happened in China on June 4, 1989&quot;?
    • shepherdjerred8 hours ago
      Do American models allow me to ask questions about piracy
      • qwerpy10 minutes ago
        I’ve been refused. I pasted a screenshot of a torrent list for a tv show. “Which one should I get, they all look the same”. “I won’t help you with copyright infringement”.<p>But the same models will tell me about the -arr stack and offer to set it up for me. You just have to frame the prompt the right way, at least for me. It can’t think it’s assisting with immediate piracy. Future theoretical piracy or for educational purposes, you’ll have better luck.
      • ATMLOTTOBEER2 hours ago
        Yeah they do it’s genuinely strange
      • fooker4 hours ago
        Yes, actually.
      • khalic5 hours ago
        Do Chinese models allow this? Tit for tat is a dumb dynamic, censoring legitimate searches is dumb no matter which “side” is doing it
    • Youden1 hour ago
      My experience so far with open-weights Chinese models (self-hosted DeepSeek, Qwen) is that they do actually answer questions about Tiananmen Square fairly objectively. The political &quot;corrections&quot; seem to be applied to domestically-hosted APIs rather than the weights themselves.
    • Retro_Dev10 hours ago
      From a far off perspective, can we really assume any model to be free of bias (political or otherwise)? I don&#x27;t know - but would like to see - how the political opinions of reddit users and blog authors are reflected in that of AI, as AI is trained on human (biased) sources.
      • nolok8 hours ago
        Realistically, they&#x27;re trained on content and thus should have the bias of their content, even without specific intent. If you train something on more chinese content it will have more bias toward it, if you train toward western centric content it will have that bias, etc ...<p>It also applies to languages, I use Qwen 27b a lot and it&#x27;s a lot smarter in english than in french.
      • Gander57396 hours ago
        <a href="https:&#x2F;&#x2F;www.economist.com&#x2F;briefing&#x2F;2026&#x2F;06&#x2F;25&#x2F;ai-models-values-are-very-different-from-most-peoples" rel="nofollow">https:&#x2F;&#x2F;www.economist.com&#x2F;briefing&#x2F;2026&#x2F;06&#x2F;25&#x2F;ai-models-valu...</a> (paywalled) might interest you.
    • Havoc1 hour ago
      Is this a task you regularly require?
    • kennywinker9 hours ago
      Is that the kind of question you ask LLMs? I use them for coding not badly summarizing history
      • omnimus7 hours ago
        Yeah I thought programmers here are &quot;apolitical&quot;. Why to care about that in a model?
        • qwerpy8 minutes ago
          Haha the programmers here are anything but apolitical. It has been an ongoing issue.<p>But I agree. If the model writes good code and makes good decisions, as long as I’m not building a video game about recent Chinese political history, I don’t care about its politics.
    • Mashimo7 hours ago
      I don&#x27;t know about qween4, but earlier chines models had a post filter on their official website, but API and selfhosted model worked fine.<p>Why do you ask? About half the time a china model gets released someone askes this. Almost as sure as a pelican.
    • cyanydeez6 hours ago
      [dead]
    • hnmodsarecucked11 hours ago
      [dead]
  • grendelt11 hours ago
    It&#x27;s great fun asking it questions like &quot;who is the Dalai Lama?&quot;, &quot;where is TSMC headquartered?&quot;, &quot;why do people post Winnie the Pooh in reference to Xi?&quot;, &quot;who controls the South China Sea?&quot;, &quot;why was Chiang Kai Shek fighting Mao?&quot;, &quot;why were students in Tiananmen Square?&quot;<p>It&#x27;ll gush about how to make moon cakes and where the tallest and longest bridges are and how to run qwen on a raspberry pi - but it can&#x27;t seem to manage some basic Wikipedia questions.
    • robertkarl11 hours ago
      I&#x27;ve done some work on this and a lot of the refusals are in the API. The weights are more permissive at these types of things. Try it again when we get some of the smaller Qwen 4 weights.
      • 0xbadcafebee10 hours ago
        You can watch it happen in DeepSeek&#x27;s chat. You&#x27;ll see it spit out the real response in the browser the first time, then the text gets cleared and it says it can&#x27;t help you with that.
    • dpkirchner11 hours ago
      Qwen3.8 flash next handles these questions just fine, what should I expect to see? It correctly reports who the Dalai Lama is, that TSLC is in Taiwan (Hsinchu), explains why people post Winnie the Pooh memes, etc.
    • lanyard-textile10 hours ago
      I suspect this happens when you run the inference in China or first party from the model maker. Out of all the open weight models I&#x27;ve used on US located hardware, they&#x27;re fine answering these.<p>Anecdotal though, N = 1.
    • incompatible10 hours ago
      I tried asking at <a href="https:&#x2F;&#x2F;chat.qwen.ai&#x2F;c&#x2F;guest" rel="nofollow">https:&#x2F;&#x2F;chat.qwen.ai&#x2F;c&#x2F;guest</a> a simple question, &quot;why did China invade Tibet?&quot; and it flashed up some transient &quot;thoughts&quot; before giving the definitive answer &quot;Oops! There was an issue connecting to Qwen3.7-Plus. Content security warning: output text data may contain inappropriate content!&quot;
    • stymaar8 hours ago
      No matter how nonsensical the laws are, you cannot expect a model maker to allow violating its country&#x27;s laws.<p>At least it doesn&#x27;t call the cops on you if you try, unlike Anthropic.
    • pmdr8 hours ago
      I&#x27;d rather deal with those refusals than Claude&#x27;s &quot;safety&quot; arrogance.
      • wolfinterval3 hours ago
        This parent comment is comical to me when Fable still switches to Opus when I ask it anything past the most basic biology in terms of weight lifting.<p>Naughty Naughty. AMPK pathway is too &quot;dangerous&quot; to talk about.
    • rolymath11 hours ago
      Chinese history is closed-weight. Chinese trade is open-weight.<p>American history is open-weight. American trade is closed-weight.
  • syntaxing11 hours ago
    Every time a major open weight is released, artificial analysis updates to a new one (v5) now. Exaggeration but some of this feels very apt (its also hilarious) <a href="https:&#x2F;&#x2F;www.reddit.com&#x2F;r&#x2F;LocalLLM&#x2F;comments&#x2F;1xkhha2&#x2F;mr_incredible_learns_the_truth_about_artificial&#x2F;" rel="nofollow">https:&#x2F;&#x2F;www.reddit.com&#x2F;r&#x2F;LocalLLM&#x2F;comments&#x2F;1xkhha2&#x2F;mr_incred...</a>
    • throwa3562622 hours ago
      <p><pre><code> Sorry, this post has been removed by the moderators of r&#x2F;LocalLLaMA. </code></pre> Has been removed, do you happen to have a copy?
      • SSLy1 hour ago
        <a href="https:&#x2F;&#x2F;x.com&#x2F;diz__zee&#x2F;status&#x2F;2116512808447672659" rel="nofollow">https:&#x2F;&#x2F;x.com&#x2F;diz__zee&#x2F;status&#x2F;2116512808447672659</a>
  • HarHarVeryFunny2 hours ago
    Did Qwen always adopt the AI butthole icon convention, or is this new?
    • Havoc1 hour ago
      Yeah always
      • Hackbraten38 minutes ago
        We’re at a point where it should be made an IETF standard.
  • trilogic7 hours ago
    Cybersecurity not mentioned. Can´t wait to test bench the open weights in cybersecurity. If is not performing well by design on it, I imagine someone have to add some additional layers XD.<p>The rest looks awesome.
  • nxtfari6 hours ago
    Posting a loss curve is an… interesting marketing approach by Qwen? I don’t know what that’s supposed to inform me about the model.
  • xiaoyu200611 hours ago
    Disappointing performance for a 2.4T model in the end of 2026.
  • Onavo9 hours ago
    The good thing is that its biotech has improved (somewhat).
  • nixosbestos12 hours ago
    Really would love if these publications have a huge, bright red bolded section &quot;How to use this right now&quot;. Is it on openrouter? Is Qwen hosting it already? Maybe I&#x27;m just ignorant?<p>EDIT: oh, I guess this isn&#x27;t open weights, so presumably I&#x27;ll have to get it through Qwen or whoever they might partner with (Bedrock?).
    • tdubey11 hours ago
      It is on openrouter: <a href="https:&#x2F;&#x2F;openrouter.ai&#x2F;qwen&#x2F;qwen4-max" rel="nofollow">https:&#x2F;&#x2F;openrouter.ai&#x2F;qwen&#x2F;qwen4-max</a><p>I would assume it will not be on Bedrock for some time, AWS has not kept up with the latest non-American models.
    • Alifatisk5 hours ago
      This has been mentioned on prior threads, you can try out all their models in chat.qwen.ai for free
  • zhuanxyz3 hours ago
    [flagged]
  • senko3 hours ago
    [dead]
  • mipo2564 hours ago
    [flagged]
  • sodowork10 hours ago
    [flagged]
  • cricketfanatic5 hours ago
    [dead]
  • Vivek-KY7 hours ago
    [flagged]
  • small_model5 hours ago
    Only 43 on AA, no where near the frontier now, looks like distillation blocking has halted their progress.
    • fancyfredbot4 hours ago
      Progress has not halted. On terminal bench 4, for example, score improved from 38.9 to 48.5.
    • testtestmetest5 hours ago
      I love how Americans are so brainwashed to think the Chinese can&#x27;t innovate without cheating, while they lap us in basically every category possible except this one verticle, that we had a 3-4 year head start on.<p>I partially wonder if the Chinese would even release a model that flat out beats Claude&#x2F;fable I say this half jokingly...<p>but I imagine there&#x27;s part of them that goes &quot;These Americans are so delusional about how language models are dangerous&#x2F;existential threats that they might kill us or hurt our economy over a chat bot&quot;<p>It could also just be they&#x27;re far more tactical. Deepseek generates a ton of revenue because its practical&#x2F;economical. You do not a mega super PhD assitant to build your 4 MAU SaaS app.<p>Anyways I hope China wins this pointless AI war, because if it does lead to innovations in medicine and science I would much rather the Chinese be the ones that discover those things than the US billionaires who&#x27;ll almost certainly try to make as much money as possible.
      • LeBit2 hours ago
        China is not better than America on any metric.<p>Okay, they have way better public transportation.<p>And they have a way better power grid.<p>And they are moving much faster on the robotics side.<p>And their EVs are killing every other nation’s EVs.<p>And they are making available LLMs that are very efficient and cost pennies to use.<p>But, other than that, they cannot operate on their own.
  • jacksonastone10 hours ago
    Anyone else worried these models have poison pills embedded in the weights? (Past examples: <a href="https:&#x2F;&#x2F;cohere.com&#x2F;blog&#x2F;concerning-behaviour-popular-open-weight-ai-model" rel="nofollow">https:&#x2F;&#x2F;cohere.com&#x2F;blog&#x2F;concerning-behaviour-popular-open-we...</a>)
    • rohansood1510 hours ago
      No technically serious person is buying the Cohere FUD you posted. I&#x27;d implore you to dig deeper and come to your own conclusion.
      • gardnr9 hours ago
        CISOs are buying it. There is no way to prove a model hasn&#x27;t been poisoned. It&#x27;s a serious implication that the behaviour can be manipulated at any time. People with heavy compliance obligations are (understandably) nervous because their customers are demanding answers. If I agree with the risk is separate from whether or not the concerns are founded in a technical understanding of the risks.
        • ctolsen6 hours ago
          So you guard the environment, like you should either way. Any model can and will eventually leak data given the opportunity, intentionally or not.
        • stymaar8 hours ago
          &gt; There is no way to prove a model hasn&#x27;t been poisoned.<p>Which is why the Cohere FUD was so efficient. That&#x27;s a bullshit claim, but a bullshit claim that cannot be proven wrong can still do massive damage.
  • bbor13 hours ago
    [flagged]
    • aiqq12 hours ago
      [dead]
    • potwinkle12 hours ago
      Another Qwen doesn&#x27;t seem all that scary, a few nation-state actors have likely exfiltrated weights of much stronger non-open models and abliterated them for use in large-scale collection on targets foreign and domestic. But god forbid someone mods a video game or jailbreaks their phone without applying for Claude Mythos or Daybreak, right?
    • zamadatix12 hours ago
      If Anthropic truly believes we are on the verge of catastrophe but nobody is listening because they&#x27;ve been crying wolf for years about it while they get rich then they should truly do something. By &quot;do something&quot; I don&#x27;t mean try again to convince the world to slow down as they IPO their bleeding edge research and solution (which is creating the doubt you mention) but just say &quot;hey, we&#x27;re so sure we&#x27;re months away from worldwide disaster that we&#x27;re shutting down our AI operations completely in hopes we can all put some attention to this before it&#x27;s too late - will you follow or work together to come up with ways to prevent this problem from happening&quot;.<p>After all, unless their goal to say &quot;I told you so but I wanted to make money more than I wanted you to believe we could stop a billion people from dying&quot; then what good is the current approach doing them?
    • pella12 hours ago
      &gt; Ah yes, an open-weights<p>This is not an open-weights modell. You have to wait for the smaller models<p><i>&quot;Smaller models in the family are coming to open source, built on the same lineage and trained with the same generation of techniques, tuned for what small models do best.&quot;</i>
      • hadlock12 hours ago
        Pretty excited about Qwen 4 Flash. We&#x27;re running qwen 3.5 35b-a3b for low effort, rapid CRM enrichment, but Q4F will be great for more complex tasks.
        • electroglyph12 hours ago
          you can use flash next right now and it&#x27;s already a massive improvement over that old 35b
      • bbor56 minutes ago
        ...really? Isn&#x27;t that <i>massive</i> change, if true? Wild! They&#x27;re burying the lede there to say the least, if that is actually a long term stance and not just the typical &quot;weights to come&quot; release cycle.