Claude Sonnet 5.5
Scoped coding & writing. Everyday coding and agents; a writing-workflow candidate.
Which reasoning effort setting?
controls how long Claude Sonnet 5.5 thinks before answering. Token prices never change; higher settings spend more and raise task cost.
Start at Max. The top setting is also the best value here. Rule: cheapest setting within 2 points of the model’s best score.
At Max, Sonnet 5.5 thinks for about 206K tokens per task, then writes about 79K more as its answer: $10.02 per task, 19.6× Low, for +28.9 points.
Output rate: $10.00/1M at every effort level. Thinking tokens are billed as output.
| Effort | Index | Per task | Thinking | All output | × Low | Time |
|---|---|---|---|---|---|---|
| Low | 25.6 | $0.51 | 8.6K | 20.9K | 1.00× | 2.6 min |
| Medium | 30.8 | $0.78 | 14.4K | 30.7K | 1.53× | 3.7 min |
| High | 41.5 | $1.64 | 31.7K | 58.4K | 3.20× | 7.1 min |
| Xhigh | 49.9 | $4.09 | 79K | 122K | 8.01× | 14.4 min |
| MaxStart here | 54.5 | $10.02 | 206K | 285K | 19.61× | 33 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 Sonnet 5.5 sits on the frontier
Intelligence Index against one 100K-input, 10K-output request.