Anthropic released Claude Sonnet 5.5 on September 28, 2026. It says the new Sonnet generates output at least 30% faster than Sonnet 5 and costs up to 30% less per task in its tests. The speed figure describes output generation, while the cost figure describes a completed task. Neither is a cut to Sonnet’s per-token price.
Sonnet 5.5 remains $2 per million input tokens, $10 per million output tokens, and $0.20 per million cache-read tokens, the same as Sonnet 5. Anthropic says it needs fewer tokens to finish the same work. Early testers also saw it batch more tool calls into fewer steps. Slack reported about 14% fewer output tokens on its offline Slackbot evaluations, alongside better results than Sonnet 5 on almost all of those tests.
Choosing between Sonnet and Opus
Anthropic positions Sonnet 5.5 for well-scoped everyday work, bug fixes, and documents. It says Opus 5.5 remains stronger on complex, open-ended tasks that require sustained judgment. Sonnet 5.5 scored 70.6% on Anthropic’s Terminal-Bench 4.0 evaluation, above the 66.4% Anthropic reports for Opus 5.5. On CursorBench 4.0, Opus scored 57.8% against Sonnet’s 55.5%. A single coding score therefore does not settle the choice for every coding workflow.
| Price per million tokens | Sonnet 5.5 | Opus 5.5 |
|---|---|---|
| Fresh input | $2 | $4 |
| Output | $10 | $20 |
| Cache write | $2.50 | $5 |
| Cache read | $0.20 | $0.20 |
At the same fresh input and output token counts, Sonnet’s listed cost is half of Opus’s. Cache reads have the same price on both models. Actual task cost also depends on how many tokens each model uses. Anthropic says Sonnet at higher effort can reach comparable performance at a similar cost to Opus; lower effort is where it describes Sonnet as the cheaper complement.
