Anthropic releases Claude Haiku 5.5, its cheapest and fastest small model
Introducing Claude Haiku 5.5: the cheapest, fastest, and most capable small model we’ve ever released. On average, it costs around 75% less to run than Claude Haiku 4.5.
Context
Builders get a small model that Anthropic says costs about 75% less to run than Haiku 4.5, and coding tools added it the same day.
Haiku is the smallest tier of Anthropic's Claude family, sitting below Sonnet and Opus. Version 5.5 is the first Haiku with an adjustable effort setting, and Artificial Analysis scored it 43 on its Intelligence Index. Cursor, Devin and OpenRouter made it available within hours of the launch.
Deep dive
Anthropic released Claude Haiku 5.5 on October 7 as the smallest and fastest model in its Claude 5.5 family. The company positions it for high-volume, cost-sensitive work such as summaries, classification and data lookups, and as a subagent that takes narrow tasks from Opus 5.5 or Sonnet 5.5 during coding work.
The main change is price. Anthropic lists $0.10 per million input tokens and $0.50 per million output tokens for prompts up to 100,000 tokens, and $0.50 and $2.50 for longer prompts. Anthropic says that works out to about 75% lower running costs on average, even after a new tokenizer that uses slightly more tokens per task.
It is also the first Haiku with an adjustable effort setting, so developers can trade cost for accuracy. In Anthropic's published results it scores 72.4% on the OSWorld 2.1 computer-use test, up from 15.7% for Haiku 4.5. Anthropic still recommends Sonnet 5.5 and Opus 5.5 for complex agentic coding.
The model is available on the Claude Platform and through AWS, Google Cloud and Microsoft Azure. Cursor, Devin and OpenRouter added it on launch day.
What to expect next
Anthropic says it is updating its SDKs to add computer use and browser use in beta, and calls Haiku 5.5 well suited to those tasks. Arena added the model to its Agent Arena with scores listed as coming soon, so independent comparisons should follow.
The open question is how much work now sent to larger models can move to a small one. Anthropic's own guidance is that Haiku 5.5 fits narrowly scoped tasks rather than complex coding.