Claude Opus 5.5 lands: beats Fable 5.1 on coding benchmarks, cuts API prices 60%
xiaohu · x · 2026-09-23
Anthropic released Claude Opus 5.5 with major upgrades to coding, vision, and writing — at much lower prices.
Key numbers
- Terminal-Bench: 66.4% vs Fable 5.1's 55.8%; CursorBench 57.8% vs 51.8%
- API pricing: $4/M input, $20/M output — 60% below Fable 5.1 ($10/$50); 40% cheaper per task than Opus 5, with 30%+ faster output
What changed
- Coding: handles long, cross-file engineering tasks with sustained autonomy
- Vision & interaction: builds complete playable works like underwater exploration games and hand-drawn animations
- Knowledge work: professional task test hits 1846 Elo (vs 1735)
- Writing: leads with the answer, less filler, follows user style rules — the official billing-bug demo shows it stating the $9.92 impact first, then explaining the code
The author warns that old habits like slicing tasks into tiny steps with constant "continue?" check-ins may now interrupt work Opus 5.5 could finish coherently.
More from coding & agent
- DigitalOcean launches Managed Agents public preview: one cloud for the whole agent loop — Scobleizer · 2026-09-23
- Engineer's verdict: AI eats people, not SWE jobs — edge-case thinkers win — kmeanskaran · 2026-09-23
- What would an ideal plugin editing environment inside ChatGPT/Codex look like? — Former_Worldliness70 · 2026-09-23
- boxd_sh praised as a major unlock for AI agents needing persistent computers — tobowers · 2026-09-23
- Open-source AI intraday trading bot ranks Nifty 50 every 15s and trades via Zerodha Kite — IndraVahan · 2026-09-23
- Debugging agent planners: which state do you actually save to reproduce a bad plan? — fishyguy3123 · 2026-09-23