fateware

@ConnorTalksAI @bridgebench GPT is good at front end, and can even pass Claude. The issue is that nobody knows how to use GPT properly. You can image generate the design and tell GPT to implement it. Something Claude cannot do.

@gitfate @bridgebench that’s fair - I assume it still should get much better at frontend though

GPT 5.6 is a leap.

@sqs @0xSero Well, the Inference provider actually decides if he wants to stream model lists with context limits, etc. So you'll have to create auto-fetching and manual import options.

@0xSero I want to add another option for any OpenAI-compatible URL, but we just need to figure out how to know which model inference to route there. Working on it.

They did it the mad men!!!!!!!

@0xSero So uhhh is AMP better than Codex?

@ConnorTalksAI @bridgebench because realistically it's not, the headline is clickbait 5.5xhigh could reach Fable performance in some tasks so 5.6 is just a step or two more basically just enough so they can pass Mythos

@bridgebench dumb question: if it is such a leap, why not brand it as an entirely new generation of models? why not call it GPT-6?

@OnlyTerp @CryptoJulzss Bro tryna slide in

@CryptoJulzss Aye a fellow pogger!

proof of gym & proof of fuel

@daniel_mac8 I think that they had months to prepare for Mythos class models, so if they don't crush Fable, then it's a complete failure by the OpenAI team.

GPT-5.6 Sol, Terra and Luna were announced by OpenAI and they only released a single popular industry benchmark. It’s either: A. GPT-5.6 crushes all benchmarks and they’re saving it for the release. B. GPT-5.6 does poorly on other benchmarks so they don’t want to advertise it. Which is it?

@orbiteditor and skills.

DeepSeek v4 Pro vs Composer 2.5 👀 Same prompt. Same design goal. The result? DeepSeek v4 Pro delivered this design for just $0.0008. The lesson isn't "always use the cheapest model." The real advantage comes from using the right open-source model with the right harness. That's where the magic happens. We're building exactly that at Orbit Editor AI, an intelligent harness that helps you get the best results from modern AI models, without wasting tokens or money. 🚀 Launching soon. Stay tuned: orbiteditorai.com #AI #OpenSourceAI #DeepSeek #LLM #WebDesign #AIAgents #BuildInPublic #OrbitEditor

For some reason I'm doing better work with GPT 5.5 medium than the GPT 5.5 xhigh It's like Xhigh thinks too much - it just completely loses itself in thought. Medium is like, okay, let me think of it, this is the implementation, done

@shaunralston Placeholder, they couldn't be bothered

How is Fable 5 going to 'reset on Friday at 5:00 AM' if the model is no longer available in subs on Tuesday? 🤔

@Alan_Earn I don't know who told you this stupidity Use image generation And it creates good UI

> GLM has no vision mode > Claude does not generate images > Gemini hallucinates too much > GPT is great at everything until you ask for frontend How?

@tekbog Just use the image generation for UI and prompt it to make it as per image... man yall cant use AI

codex frontend skills

@kr0der You do realize 5.5 is just being used wrong when doing frontend, it literally has image generation capabilities. Use them to build you a good UI and then prompt GPT to make it as per image. You can simplify this by using this skill github.com/Master0fFate/j…

5.6 Sol Ultra is about to be insane man, so much hype around it if it’s good at frontend/creative stuff it’s gg

@MarcosHernanz Codex cuz you will get 5.6 soon lol.

- Codex: No Fable 5 - Cursor: No caching with BYOK - Conductor: No done notifications - T3 Code: No forks - OpenCode: No worktrees What coding app do I even use? 😭

@DavidOndrej1 You can also check my skillset + agents md i run daily. 🙃 github.com/Master0fFate/j…

im releasing all my agent skills to the public this is hundreds of hours of trial & error every single global .agents skill i have go grab it. it's free.

real and true

@thsottiaux Performance, which still needs a bit of tuning btw

What is something that you feel is surprising that Codex still can't do well and we should have gotten right a while ago?

@Aditya_181105 GUI ofc

AI is splitting into two worlds: GUI users Vs CLI users Which one wins long-term?

@SimonHoiberg @botensten asked the same question - 3 months ago. Now he is livestreaming his journey to becoming "technologically-independant".

"Just use Vercel." "Just use Supabase." "Just use Clerk." Cool. Now your auth, database, and deployment are owned by 3 different companies who can change pricing whenever they want. And the rest of your product is wrapping OpenAI. At some point you have to ask yourself: what do I actually own here?

how it feels having a sensitive heart with an idgaf mindset