Claude Haiku 5.5 is Anthropic's new small model
Haiku 5.5 is the fastest, cheapest small Claude yet, with a 1 million token context window and adaptive thinking.
Official source (opens in a new tab)
Anthropic's product page, dated 7 October 2026, calls Claude Haiku 5.5 the cheapest, fastest, and most capable small model the company has released. The Claude Platform release notes from the same day give the API details: model id claude-haiku-5-5, a 1 million token context window, 128k maximum output tokens, and adaptive thinking with the effort parameter. It is on the Claude API, Amazon Bedrock, Google Cloud, and Microsoft Foundry.
The same notes say code written for Haiku 4.5 can break. Manual extended thinking, temperature, top_p, top_k, and assistant prefill can return a 400 error. Anthropic also cut the price of prompt-cache reads on Claude Sonnet 5.5 that day, from $0.20 to $0.10 per million tokens.
Who it's for: high-volume, latency-sensitive work and subagents, not the model you open for a long draft.
Fits: Best for Writing only as the cheap sibling of the Claude product. The agent use is closer to Best Agent Builder.