Claude Opus 5.5 tops the Coding Agent Index at 66, but cost per task jumps 21%
ArtificialAnlys · x · 2026-09-24
Artificial Analysis reports Claude Opus 5.5 is the new #1 in the Coding Agent Index, scoring 66 at max effort in Claude Code — the highest score they have measured, up 6 points over Opus 5 (60) and 4 over Claude Fable 5.1 (62).
Gains across all three evaluations:
- Terminal-Bench 4.0: 54.5% → 63.1% (+8.6 points, largest gain)
- DeepSWE v1.1: 62.5% → 68.4%
- SWE-Atlas-QnA: 62.1% → 66.4%
Pricing and cost: Anthropic cut Opus pricing to $4/$20 per million input/output tokens (from $5/$25), and cache reads to $0.20 from $0.50. But with substantially more tokens (15.6M per task vs 11.4M, and 2.4× output tokens), cost per task is $13.04, up 21% from Opus 5's $10.79.
Even so, Opus 5.5 extends the score-vs-cost Pareto frontier: no lower-cost model in the comparison matches its score.
More from coding & agent
- Meta's Muse Connector Platform draws 2,000+ developer submissions days after launch — Rasmic · 2026-09-24
- Why AI-made dev tools beat one-shot game generation: freedom of process — eschadiol · 2026-09-24
- RealSense VP on AgenticROS: Letting AI Agents Directly Control Physical Robots — chrismatthieu · 2026-09-24
- AWS API Gateway's Hard 10MB Upload Limit and the Presigned URL Fix — _jaydeepkarale · 2026-09-24
- Open-source Pragma gives coding agents a terminal-first workspace with Git worktrees — tech_w0rld · 2026-09-24
- Dev builds dense task annotation system with GPT-6 Astra, ships it as an LLM skill — chris_j_paxton · 2026-09-24