Seems like a whole lot of nothing for the average user. They have really lost the plot.
For those who had access, how does it compare IRL with GLM 5.3 ? iirc both models are similar in terms of benchmarks ?
may have been rushed by hugging face being unable to use claude to debug or fix their breach, because there isnt detail on new guardrails put into place to protect against anthropic's initial concerns of wide distribution?<p>separately, it is interesting they are adding an oem type offer:<p>>We’re working with our cybersecurity technology and services partners to integrate Claude Mythos 5 into the products and services defenders already use to secure their software.<p>anyone know the shape of those contracts, e.g. pure tokens/usage or more of a traditional licensing or oem type structure?
> Bringing the cybersecurity capabilities of Claude Mythos 5 to more defenders<p>This dance is quite annoying. If Anthropic release a paid model, the user should control what it does and doesn't do. The other day I had a security incident that required urgent response. ChatGPT and Claude were utterly useless (I quickly attempted to get access to the former's advanced security capabilities but was met with a form I could complete - far too slow for a time-sensitive problem like a security incident!). So I used grok and it helped! I am told Kimi also helps in such cases although haven't tried it yet.
This week I found two issues in my company codebase. After finding them I told a claude session about one and asked for a quick proof of concept demo of the exploit. It refused, including refusing simple things in the same session afterwards.<p>Meanwhile same model in a new tab, say I need help creating a page that hits an endpoint with a special payload and it does the same things that were too dangerous in the previous tab...
I had that when opus 4.8 was first out. It repeated refused to make a PoC for an issue it suspected.<p>I eventually gave up and just asked it to fix the issue. The first thing it did? Write a PoC to verify the issue was still valid...
Just use a EU or US hosted Asian Opensource model. They will spit out everything.
In a world where things such as GLM-5.3, DeepSeek-V4-Pro-0813, and Kimi-K3 exists, this is a bit laughable.<p>Anthropic needs some model with a fancy name so they can pretend for another while that their model is so powerful it will destroy the world if released. I propose Claude Legend 6.
China will beat them to general availability.<p>Anthropic is overconfident, and it's going to bite them in the ass.<p>Dario deserves it for being so hypocritical, fear mongering, and ladder pulling. Fastest hero to villain arc in recent memory.