Claude Haiku 5.5 released
On 7 October 2026 Anthropic released Claude Haiku 5.5, which it calls its cheapest, fastest and most capable small model. It says the model costs around 75% less to run than Haiku 4.5, on average; the list price is $0.10 and $0.50 per million input and output tokens for prompts up to 100,000 tokens, against $1 and $5 for Haiku 4.5. It was available the same day on the Claude Platform and on AWS, Google Cloud and Microsoft Azure.
Why it matters
Anthropic's small-model tier now costs about a tenth of its predecessor for prompts up to 100,000 tokens, which the company says made up about 90% of requests to Haiku 4.5. It positions the model for summaries, classification and database queries and as a subagent beside Opus 5.5 and Sonnet 5.5, and says in plain words that the larger models remain the better choice for complex coding. It is also the first Haiku-class model with an adjustable effort setting.
What the announcement says. Claude Haiku 5.5 (model id claude-haiku-5-5) is described as designed for high-volume, cost-sensitive work: summaries, compaction, database queries and classification; as a subagent beside Opus 5.5 and Sonnet 5.5 on coding work; and for speed-sensitive tasks such as live customer support and browser use. It is the first Haiku-class model with an adjustable effort setting. It is available on the Claude Platform, Amazon Web Services, Google Cloud and Microsoft Azure. Price. Per million tokens, for prompts up to and over 100,000 tokens: input $0.10 and $0.50, output $0.50 and $2.50, cache reads $0.01 and $0.05, cache writes $0.125 and $0.625; Haiku 4.5 was $1 for input and $5 for output. The page says the average cost to run falls by around 75%; a footnote explains that the price is 90% lower for requests up to 100,000 tokens and 50% lower above that, that 90% of requests to Haiku 4.5 were in the first group, and that the calculation allows for a new tokenizer that uses slightly more tokens per task. The same day Anthropic halved Sonnet 5.5's cache-read price (from $0.20 to $0.10 per million tokens), which it says makes Sonnet 5.5 around 20% cheaper on most agentic work, and announced a monthly API credit for Max and Team subscribers (Max 5x $100, Max 20x $200, Team up to $500 pooled). Anthropic's own figures. Haiku 5.5 against Haiku 4.5, GPT-6 Luna and Sonnet 5.5 (for reference): GDPval-AA v2.1 1620 Elo against 735, 1437 and 1840; OSWorld 2.1, offline subset, 72.4% against 15.7%, 48.9% and 83.9%; Humanity's Last Exam without tools 45.9% against 10.2% and 56.9%; Terminal-Bench 4.0 39.2% against 0.0%, 16.4% and 70.6%. The page says Sonnet 5.5 and Opus 5.5 remain better choices for complex agentic coding and that Haiku 5.5 suits narrowly scoped tasks. Safeguards. The page reports far fewer instances of misaligned behaviour than Haiku 4.5 in its alignment evaluations. Its cybersecurity safeguards are more restrictive than Haiku 4.5's but somewhat less restrictive than those on other recent models: they allow a wider range of defensive tasks than Sonnet 5.5's, but still block penetration testing. Its biology safeguards are the same as for Sonnet 5, Sonnet 5.5 and Opus 5. What the record does not claim. All benchmark figures, the price comparison and the safety statements are Anthropic's own; the page names Artificial Analysis as the source of GDPval-AA, but no independent assessment of Haiku 5.5 was read. The customer quotations on the page were not used.