Anthropic has released Claude Sonnet 5.5, positioning the model as a faster and more efficient successor to Sonnet 5 without changing its token prices. The model costs $2 per million input tokens and $10 per million output tokens, the same pricing listed for Sonnet 5.

Anthropic reports that Sonnet 5.5 runs 30%+ faster. More efficient tool-call batching can also reduce the cost of a task by up to 30%, according to the release details. The improvement is therefore tied not only to the model's stated speed, but also to how it handles tool calls during a task.

Benchmark performance

The largest reported gain comes on Terminal-Bench 4.0, where Sonnet 5.5 scores 70.6%. Sonnet 5 is listed at 10.3% on the same benchmark.

The new model also records 80.1% on OSWorld 2.1. On GDPval-AA v2.1, it reaches 1,844, narrowly below the 1,846 score listed for Opus 5.5.

These figures place Sonnet 5.5 close to Opus 5.5 on GDPval-AA v2.1 while retaining the pricing associated with the Sonnet line. The benchmark results supplied for Terminal-Bench 4.0 show the clearest comparison with the previous Sonnet release.

Cloud availability

Claude Sonnet 5.5 is available across AWS, Google Cloud and Azure. Its identifier on those platforms is claude-sonnet-5-5.

The cloud availability gives customers using those services a direct model identifier for accessing the release. The supplied details do not include separate pricing or rollout conditions for each cloud platform.

Anthropic has also teased a Haiku 5..., although no further details about that model are included here.

Pricing and practical impact

For customers comparing model costs, the listed token rates remain unchanged from Sonnet 5: $2 per million input tokens and $10 per million output tokens. The reported up-to-30% reduction applies per task and is attributed to more efficient tool-call batching rather than a lower published token price.

The release combines that cost claim with a 30%+ speed improvement and higher results on the listed benchmarks. The most notable reported comparison is Terminal-Bench 4.0, where Sonnet 5.5's 70.6% score is substantially above Sonnet 5's 10.3%.

Conclusion

Claude Sonnet 5.5 brings faster operation, lower reported per-task costs and stronger benchmark results while keeping Sonnet 5's listed token pricing. It is available through AWS, Google Cloud and Azure under claude-sonnet-5-5.

Frequently Asked Questions

Q. What is Claude Sonnet 5.5?

Claude Sonnet 5.5 is Anthropic's newly released model and a successor to Sonnet 5.

Q. How much does Claude Sonnet 5.5 cost?

It is priced at $2 per million input tokens and $10 per million output tokens.

Q. How much faster is Sonnet 5.5?

Anthropic reports that it runs 30%+ faster than Sonnet 5.

Q. What is its Terminal-Bench 4.0 score?

Sonnet 5.5 scores 70.6% on Terminal-Bench 4.0.

Q. Where is Claude Sonnet 5.5 available?

It is available across AWS, Google Cloud and Azure under the identifier claude-sonnet-5-5.