Anthropic releases Claude Haiku 5.5, a small model it says costs about 75% less to run

Anthropic released Claude Haiku 5.5 on 7 October 2026, calling it the cheapest, fastest and most capable small model it has shipped.

Key facts

• Purpose: high-volume, speed-sensitive tasks — summaries, classification, database queries, live customer support, browser use, and acting as a sub-agent for the larger Opus 5.5 and Sonnet 5.5 models.
• Price: $0.10 per million input tokens and $0.50 per million output tokens for prompts up to 100,000 tokens (Haiku 4.5: $1 and $5). Anthropic says that on average the model costs around 75% less to run, partly offset by a new tokenizer that uses more tokens for the same text.
• Specs: a 1 million-token context window, up to 128,000 output tokens, and — for the first time in a Haiku model — an adjustable effort setting.
• Benchmarks (company-reported): on the OSWorld 2.1 computer-use test (offline subset), Haiku 5.5 scored 72.4% versus 15.7% for Haiku 4.5. Anthropic says Sonnet 5.5 and Opus 5.5 remain better for complex agentic coding.
• Availability: Claude API (claude-haiku-5-5), Amazon Bedrock, Google Cloud and Microsoft’s platform.

Other changes

• The price of cache reads on Sonnet 5.5 is cut by 50%, from $0.20 to $0.10 per million tokens.
• Max 5x subscribers get $100 a month in API credits, Max 20x get $200, and Team plans up to $500 pooled.

All performance figures come from Anthropic and its customers; independent evaluations will follow.