Model Details
Claude Opus 5.5
Quick answer: Claude Opus 5.5 is the first model in Anthropic's new Claude 5.5 family, released September 22, 2026. Anthropic says it performs at the level of Claude Fable 5.1 on most work while costing 40% less to run than Claude Opus 5 — input pricing drops from $5 to $4 per Mtok, output from $25 to $20 per Mtok.
At a Glance
- Where it leads: Terminal-Bench 4.0 (66.4%), GDPVal-AA v2.1 (1,846 Elo), Humanity's Last Exam with tools (67.7%) — all above Claude Fable 5.1 and Claude Opus 5 in Anthropic's own comparison.
- Where it lags: Terminal-Bench-Science 0.1 (58.7%), below GPT-6 Astra's 64.6% on the same eval.
What Claude Opus 5.5 Is
Claude Opus 5.5 is Anthropic's cost-reduced successor to Claude Opus 5, aimed at matching Claude Fable 5.1-level performance on most tasks at a lower price point. Cache-read pricing also drops 60% (from $0.50 to $0.20 per Mtok), and a faster mode is available at $8/$40 per Mtok.
Specifications
| Organization | Anthropic |
| License | Proprietary (API only) |
| Release date | September 22, 2026 |
| Modality | Text |
| Knowledge cutoff | Not disclosed |
Public Benchmark Scores
| Benchmark | Score | Source | Date |
|---|---|---|---|
| Terminal-Bench 4.0 | 66.4% (xhigh effort) | Anthropic | Sep 2026 |
| FrontierCode v1.1 (Main) | 54.4% | Anthropic | Sep 2026 |
| GDPVal-AA v2 | 1,846 Elo (v2.1-anchored, max effort) | Anthropic | Sep 2026 |
| Humanity's Last Exam | 67.7% (with tools, max effort) | Anthropic | Sep 2026 |
| Terminal-Bench-Science 0.1 | 58.7% (max effort) | Anthropic | Sep 2026 |
| OSWorld 2.0 | 81.8% (partial) | Anthropic | Sep 2026 |
Scores as reported in Anthropic's own Claude Opus 5.5 announcement (anthropic.com/claude-opus-5-5, September 2026); not independently verified by Benchgen.