Follow-up: Opus 5.5 excels at building tools to improve its own work
victormustar · x · 2026-09-23
In a follow-up to his Boeing 747 benchmark run, the author calls Opus 5.5 "a beast," highlighting its ability to build the right tools for the job and continuously improve its own output. Full details are in the original thread's shared prompt and run.
Related event: Opus 5.5 aces Boeing 747 benchmark, praised for building its own tools(2 posts)→
More from Models
- GPT-6 verdicts split: Sol disappoints, Luna surprises, and Opus 5.5 wins users back — xeophon · 2026-09-23
- Meme Roasts Google for Endless Flash Models Instead of Competing at the Top — RichardMortis12 · 2026-09-23
- Decision-only model Jev beats frontier Gemini on 1,759 real decisions: 5x faster, 25x cheaper — Humble_News_6994 · 2026-09-23
- 40% Opus price cut and AGENTS.md support: Anthropic's four plays to win back developers — iannuttall · 2026-09-23
- Anthropic and OpenAI price cuts show open-source models are winning — bindureddy · 2026-09-23
- Apple AI is still unusable: asks for today's date, confidently answers Oct 10, 2024 — bernard_hossmoto · 2026-09-23