fly³ retweeted
I feel like I haven't said this enough lately, or even at all, and I need to say it.
Opus 4.8 is a good model, and Claude Code is a very solid harness. What @bcherny and team are doing is exceptional, and nobody should be sleeping on this combo.
While I do talk about Codex and GPT 5.5, which has been my daily driver for the last couple of months, that doesn't mean I don't like Claude Code and Opus 4.8.
I use both. For whatever reason I just tweet about Codex a lot more, and that's probably because I like their Desktop app a lot, this is probably the biggest gap rn between the two.
Claude Code is very comparable to Codex when you compare CLI to CLI, but app to app, Codex is significantly better than Claude Code in the Claude app. This is where @thsottiaux and team have really shown some serious innovation and focus.
But I'm hearing people diss Opus 4.8 way too much lately. It is a very good model, and while I do like GPT 5.5 more, that doesn't mean 4.8 is bad.
Still have plenty of love for you Anthropic 🫶
fly³ retweeted
We never had more npm downloads than this week on @openclaw - comined with Docker, GitHub, company-internal deployments and the numerous forks, real number is more in the 10-20 million downloads/week.
Most people are still treating AI like a search engine.
The ones who figure out how to use it as a thinking partner — they're going to have a serious edge in the next 5 years.
The gap is already forming.
1
91
fly³ retweeted
You can now use your @grok or X Premium subscription in @opencode.
Use the model powering Grok Build for high speed and codebase intelligence.
x.ai/news/grok-opencode
fly³ retweeted
OpenAI frontier models and Codex are now generally available on AWS, giving enterprises a new way to build on Amazon Bedrock with OpenAI through the security, compliance, and governance workflows they already use.
This is also the beginning of a broader expansion of OpenAI capabilities on AWS, including future availability for cybersecurity capabilities like Daybreak.
openai.com/index/openai-fron…
fly³ retweeted
I'm running a gazillion codex and 100s of vms.
Only way i can work now, each tmux pane is another tmux to a remote m4 mac mini via mosh with 6x4.
Then each one of those codex sessions is farming out all its testing on mac, windows, docker on AWS and Azure spot instances via crabbox.
fly³ retweeted
I want to share some unglamorous work I’m proud of.
Starting in late March and landing around the Apr 7 OpenClaw release, we intentionally cut aggregate OpenRouter token usage by ~35%, down to ~400B tokens.
This wasn’t one trick. It was good engineering across paths agents hit all day: oversized tool results, compaction/overflow recovery, background and subagent context, and prompt caching.
My part was mostly the tool-result / overflow path. Vincent worked on cache boundaries and fingerprints. Boris fixed deterministic tool ordering and cache-preserving compaction. Ayaan helped with subagent light context and nearby context-shaping fixes.
Relevant commits in reply.
fly³ retweeted
what pushed us to finally implement it is heavier use of worktrees
now that i have sessions going across different worktrees very annoying to find them and open my editor in the right spot
we'll ship the worktree feature out from the flag next week
fly³ retweeted
imagine a benchmark of two editors
one opens a file 10x faster! wow it must be better. but oh wait they just disabled syntax highlighting
real products have to do things that make it worse on benchmarks but better in practice
