Claude has announced Claude Opus 5.5, the first model in its Claude 5.5 family. The company claims it performs at roughly the level of Claude Fable 5.1 for most tasks while costing 40% less to run than Opus 5.
Claude positions Opus 5.5 as an improvement over Opus 5 in agentic coding, computer use and knowledge work. Its published benchmarks show gains across several tests, although most results use adaptive thinking at maximum effort and should be read in that context.
| Benchmark | Opus 5.5 | Opus 5 | Fable 5.1 |
|---|---|---|---|
| Terminal-Bench 4.0 | 66.4% | 52.3% | 55.8% |
| FrontierCode v1.1 Main | 54.4% | 48.0% | 50.3% |
| CursorBench 4.0 | 57.8% | 46.6% | 51.8% |
| GDPval-AA v2.1 | 1,846 | 1,708 | 1,735 |
| AutomationBench | 40.0% | 26.9% | 31.4% |
| Humanity’s Last Exam with tools | 67.7% | 63.6% | 65.6% |
Opus 5.5 also scored 81.8% on OSWorld 2.0 with partial computer-use credit, compared with 80.7% for Fable 5.1 and 74.0% for Opus 5. On Chartography’s visual chart-recognition test with tools, it reached 89%, narrowly above Fable 5.1’s 88.4%.
The comparisons are not uniform. GPT-6 Astra scored higher on AutomationBench, at 41.4%, and Terminal-Bench-Science, at 64.6%, while Opus 5.5 reached 40% and 58.7% respectively. Claude’s benchmark notes also indicate that Opus 5.5 was evaluated with production safeguards enabled, and that the Terminal-Bench results use high-effort settings.
Claude claims the new model generates output more than 30% faster than Opus 5. The company also describes changes to its communication style, including putting important information first and following user-provided writing rules more consistently.
Pricing is lower across each listed token category. Opus 5.5 costs $4 per million input tokens and $20 per million output tokens, compared with $5 and $25 for Opus 5. Cache reads cost $0.20 per million tokens, down from $0.50, while cache writes cost $5 per million tokens, compared with $6.25.
Claude states that Opus 5.5 underwent external evaluation before release, including assessments by METR and Frontier Design. The company also claims the model achieved its strongest score so far on its most comprehensive alignment test, without providing a score in the announcement.
Alongside the model launch, Claude is increasing five-hour usage limits for Pro, Max and Team plans. Subscription users will also receive a rate-limit reset that can be saved for later use.
Source: Claude on X
