CLAUDE HAIKU 5.5 IS ANTHROPIC'S FASTEST MODEL YET

Anthropic released Claude Haiku 5.5 on October 7, calling it the cheapest, fastest and most capable small model it has ever shipped. It joins Opus 5.5 and Sonnet 5.5, and Anthropic says it is the fastest model it has released to date.

What Haiku 5.5 Is Built For

Anthropic aims it at high volume work that has to be cheap, like summaries, compactions, database queries and classification. It is also pitched as a subagent that runs underneath Opus 5.5 and Sonnet 5.5 on coding jobs, and for speed sensitive work like live customer support and browser use.

It is the first Haiku model with an adjustable effort setting, so developers can dial cost against intelligence on one model. It has a 1 million token context window and up to 128,000 tokens of output.

The Benchmark Jump From Haiku 4.5

The gap over the last Haiku is wide. On OSWorld 2.1, which measures an agent operating a real computer, Haiku 5.5 scored 72.4 percent against 15.7 percent for Haiku 4.5. On the agentic coding benchmark Terminal Bench 4.0 it scored 39.2 percent, where Haiku 4.5 scored zero. On Humanity's Last Exam it reached 45.9 percent without tools. Sonnet 5.5 still leads on all three, at 83.9 percent on OSWorld and 70.6 percent on Terminal Bench.

The Cost Story

Anthropic says Haiku 5.5 costs about 75 percent less to run than Haiku 4.5 on average. Its API pricing starts at $0.10 per million input tokens and $0.50 per million output tokens.

The launch came with two more changes. Anthropic halved the price of Sonnet 5.5 cache reads, which it says makes Sonnet about 20 percent cheaper on most agentic work, and it added a monthly API credit for Claude Max and Team subscribers to build on the Claude Platform.

It is the latest price move in a lineup that has kept getting cheaper at the top too, after Opus 5 arrived at half the price of Fable 5. For where the frontier sits now, see our look at the Fable 5.1 benchmarks. Haiku 5.5 is available now in the API as claude-haiku-5-5.