Claude Opus 5.5
Default starting point. Anthropic’s recommended default for most work, including long-running coding and knowledge work.
Which reasoning effort setting?
controls how long Claude Opus 5.5 thinks before answering. Token prices never change; higher settings spend more and raise task cost.
Start at Xhigh. It scores 57.4 against 59.1 at the top setting, for 40% less per benchmark task. Rule: cheapest setting within 2 points of the model’s best score.
At Xhigh, Opus 5.5 thinks for about 63.8K tokens per task, then writes about 39.6K more as its answer: $6.05 per task, 6.2× Low, for +22.6 points.
Output rate: $20.00/1M at every effort level. Thinking tokens are billed as output.
| Effort | Index | Per task | Thinking | All output | × Low | Time |
|---|---|---|---|---|---|---|
| Low | 34.8 | $0.98 | 6.9K | 19.1K | 1.00× | 3.1 min |
| Medium | 48.2 | $2.45 | 20.7K | 43.8K | 2.51× | 6.8 min |
| High | 51.8 | $3.47 | 31.6K | 60.1K | 3.55× | 9.3 min |
| XhighStart here | 57.4 | $6.05 | 63.8K | 103K | 6.20× | 16 min |
| Max | 59.1 | $10.04 | 120K | 173K | 10.28× | 25.6 min |
Per task of the Artificial Analysis Intelligence Index: the average over its nine evaluations, counting input, cache, and output tokens at list prices. These are long, agent-style benchmark tasks, much longer than a chat message.
Where Claude Opus 5.5 sits on the frontier
Intelligence Index against one 100K-input, 10K-output request.