Just put the finishing touches on the Landing Page for Sotto. Fable 5.1 is still the king at frontend design in my opinion.

Small change, but "/claude-api prompt-audit" is now also "/checkup prompt-audit" The name made it sound API-only, even though it always worked on your Claude Code setup. It checks your CLAUDE.md, skills and agents for instructions your model doesn't need anymore. Really useful!

Decided to put together a Model Tier List based on the Models I have used. This is not a pure intelligence or cost ranking. It combines how useful the models are for different workflows. Agree or Disagree? Happy to debate lol

The right way to give feedback about model changes /feedback and post about issue with actual examples. The wrong way, vague post with no example on X for engagement.

Just put the finishing touches on the Landing Page for Sotto. Fable 5.1 is still the king at frontend design in my opinion.

One thing I really appreciate about OpenAI models is how persistent they are when chasing down a bug or issue. They will run every test, throw every possible edge case at a wall and find the issue. I have not found that persistence to find an issue in any other model.

2-3 months from now where will Anthropic and OpenAI be? I honestly think SpaceXAI needs to stop chasing the frontier and focus on a fast efficient engineering model. That can be used by other larger models

Fable 5.1 vs. Opus 5.5 Opus 5.5 is an incredible model. It’s a leap forward in almost every category. It’s cost efficient, nice to talk to, decent at coding, and amazing at 3D modeling. It’s my new daily workhorse. With Opus 5.5 scoring significantly higher than Fable 5.1 on benchmarks and costing significantly less, is there still a reason to use Fable? I’d argue yes, but for a much narrower set of use cases. I’m reaching for it significantly less. Fable 5.1 is still my go to when I want something designed with that extra touch of taste and polish I haven’t found in any other model. I also tend to reach for it when I’m writing user facing code. But increasingly, I’m having Fable orchestrate Opus 5.5 rather than do all the work itself. Anthropic has truly done something special here. With Fable 5.1, we have a model that’s extremely good at working through the most difficult problems. Now, with Opus 5.5, we also have a cost efficient daily workhorse that can handle almost anything. The new $200 subscription finally feels like it might actually cover a full week of real world work.

I am detecting some shady plays on Claude team usage. Some sessions I have huge usage some get wiped in 1 hour. There is something inconsistent. [team premium]

Opus 5.5 is a pretty good model.

Insane

Are you still using Fable 5.1 now that Opus 5.5 is out?

The more I use gpt-6-Sol the more disappointed I get. I initially was happy with some of the results it gave me but then I started noticing some regressions so I had Opus 5.5 go over its changes and there were a decent amount of issues. It especially struggled with creating an excel file formatted properly, which I haven’t had that issue in a long time. Looking like this might actually be a flop. Unless something better comes out soon I might just be getting a second Claude account.

Opus 5.5 is AGI. Here is the BO2 Zombie game it made. This was by no means a one-shot and still has some rough edges, but this game feels as close the original as I remember. Insane. Everything in this other than Audio was generated by Opus 5.5. This took about 12 hours and cost more than I would care to say lol