I'm burning 50% less weekly Claude Code limits now.
1. Install OpenAI Codex plugin inside Claude Code
2. Then use:
> Claude Fable 5 high for planning
> Codex 5.5 xhigh for execution (using Codex plan, no API)
> Claude Fable 5 max for review
That's it. It works.
@freddier honestly, codex desktop is good enough that i could use either, but claude is still a bit behind
but the main advantage of the cli is customization. with cmux you can build a workspace that fits exactly how you like to work
@andrewchen Infrastructure that enables products to communicate with users’ AI assistants like Hermes or Openclaw for real-time, continuous personalization.
Today apps repeatedly relearn users through repetitive onboarding flows while still failing to adapt as users evolve over time.
GPT-Image-2 had a 93% win rate in Image Arena.
Arena rankings come from blind, pairwise battles where voters pick between two anonymized image outputs for the same prompt.
GPT-Image-2 from @OpenAI was preferred 93% of the time, resulting in a record-breaking +242 point leap ahead of the competition on Arena's Text-to-Image leaderboard.
@LangChain@vercel for those looking for more plug-and-play options, @linear agents integrations site has a nice list of tools in this space
https://t.co/hdVMygSN9x
lately i’ve been getting into cloud coding agents
big companies are building in-house systems (stripe minions, ramp inspect) that generate thousands of PRs per week
if you want to set one up, here are a few oss options:
@LangChain open-swe → framework
@vercel open-agents → template
@colemurray background-agents → full system
@sregg@callstackio was planning to compare these soon too, super helpful, thanks for sharing
curious what you think about argent when you try it
expo probably will jump in soon too, evan’s been playing with embedding the simulator into most coding agents
https://t.co/4udWzsTV3n
@colemurray@lucasmeijer curious what pushed you to build this instead of using other open source alternatives like langchain open-swe or vercel open-agents
been researching this space recently, would love your thoughts
@bcherny recap only works in new conversations?
tried to use it on old conversations and got "No recap available — needs at least one completed turn, or generation failed"