Anthropic launches Claude Sonnet 5.5 with 1M context

Anthropic launched Claude Sonnet 5.5 on September 28, 2026, according to its release notes, and later documented the model’s pricing, limits and platform availability in its product docs. The company positions Sonnet 5.5 as the second model in the Claude 5.5 family, following Claude Opus 5.5 on September 22, 2026 and ahead of Claude Haiku 5.5 on October 7, 2026.

The release matters because Sonnet 5.5 is designed to sit in the middle of Anthropic’s model lineup while still carrying enterprise-grade scale. Anthropic says the model is built for coding, agent and enterprise workloads, including code generation, data analysis, content creation, visual understanding and agentic tool use. The company also describes it as the best combination of speed and intelligence in its Sonnet line.

What does Claude Sonnet 5.5 offer developers?

Claude Sonnet 5.5 gives developers a 1 million-token context window and a 128,000-token maximum output, according to Anthropic’s model overview. In practical terms, the larger context window lets teams feed in much longer documents, codebases or conversation histories before the model has to drop earlier material. The output cap also allows for long responses, including extended code generation or analysis.

Anthropic says the model uses the same tokenizer as Claude Sonnet 5, so the same text produces the same token counts. The company also says adaptive thinking is on by default, with the default effort set to high on the Claude API. For users who want less up-front reasoning, the lowest thinking setting is between_tools, which turns off up-front thinking.

The model supports per-message effort, mid-conversation system messages, mid-conversation tool changes, prompt caching, batch processing, the Files API, PDF support, vision, and server-side and client-side tools. Anthropic also says Sonnet 5.5 supports computer use through the computer_toolset_20260801 toolset.

How much does Claude Sonnet 5.5 cost?

Anthropic lists Claude Sonnet 5.5 at $2 per million input tokens and $10 per million output tokens. It also says the model’s prompt-caching prices are $2.50 per million tokens for a five-minute cache write, $4 per million for a one-hour cache write and $0.10 per million for cache reads.

The company says the model has the same prices as Claude Sonnet 5 except for prompt cache reads, which cost half as much on Sonnet 5.5 at $0.10 per million tokens. Anthropic also says the Batch API offers a 50% discount on input and output pricing.

For some batch workflows, Anthropic says Sonnet 5.5 supports up to 300,000 output tokens on the Message Batches API with a beta header. The minimum cacheable prompt length is 512 tokens.

Where is Claude Sonnet 5.5 available?

Anthropic says Claude Sonnet 5.5 is available to all Claude API customers under the identifier claude-sonnet-5-5, and it is also available through Amazon Bedrock, Google Cloud and Microsoft Foundry. The overview page also lists Claude Platform on AWS among the supported platforms.

Anthropic provides platform-specific model IDs as well. For Amazon Bedrock, the model ID is anthropic.claude-sonnet-5-5, and for Microsoft Foundry it is claude-sonnet-5-5. The company says the model launched on September 28, 2026 and will not retire sooner than September 28, 2027.

Anthropic lists June 2026 as both the reliable knowledge cutoff and training data cutoff for Sonnet 5.5. It also says setting temperature, top_p or top_k returns a 400 error.

What guardrails and limits does Anthropic disclose?

Anthropic says Sonnet 5.5 can decline requests in five stop_details categories: cyber, bio, frontier_llm, reasoning_extraction and general_harms. If a request is declined, the API returns HTTP 200 with stop_reason set to refusal and a stop_details object naming the policy area.

The company also says the model does not support forced tool use using tool_choice type any or tool. That matters for builders integrating the model into agentic systems, because it limits certain ways of constraining tool execution.

Anthropic’s newsroom has also described the broader Claude 5.5 family as cheaper, faster or more capable than prior models, but the snippet available there does not fully establish pricing, availability or benchmark details for those claims. For Sonnet 5.5 specifically, Anthropic says it runs 30%+ faster than Claude Sonnet 5 and costs up to 30% less for most work.