[revolvingthrow]

Benchmarks look really promising. Suspiciously good, even. I guess we’ll see soon enough.

My question to previewers: how are the guardrails for random joe that wasn’t personally blessed by the ai pope to access the non-nerfed model? Fable is a nightmare in this regard, but I’m not sure whether 5.6 also gets a critical side-eye from the gubmint when you ask it to fix bugs in your code (you filthy hacker, you).

[cmrdporcupine]

I almost immediately ran into "This request requires additional safety checks, which can take extra time. Hang tight or retry with a faster model for a quicker response, though it may be less capable of handling complex requests."

Which is something I've never seen with codex before, and I wasn't doing anything funky. Just writing CUDA kernels and benchmarks for them.

[matheusmoreira]

Is it actually usable though? Because the Fable situation is just obnoxious. If OpenAI's Fable equivalent is actually usable, I'll cancel my Anthropic subscription on the spot.

[cmrdporcupine]

Yes it's usable, however it does make it difficult to just do /goal and walk away if this is happening periodically. I don't do that often, but for well defined things I sometimes do.

[matheusmoreira]

Great. I usually babysit the agents so I think it won't be a problem. So long as there's no obnoxious downgrading going on.

That settles it. Anthropic has until the end of this month to get rid of Fable's "safety" bullshit. If they don't, I will not renew my subscription.

[alex0015]

I was getting that regularly last week with regular 5.5 medium on the plus plan. I was doing benchmarking for a photo editor in Swift.

[cmrdporcupine]

Interesting. Never seen it before. Now it's just constant and there's nothing sensitive about this work.

[Donald]

I'm getting the same message doing WebGL shader work.

[cmrdporcupine]

It (Sol, on high) does seem actually really quite competent at this work though (GPU programming). Much more so than the attempts I made with 5.5 earlier today.