Skip to content

[ commentary ]

Haiku 5.5 cuts the price of a short request by 90%

Anthropic's Haiku 5.5 costs 90% less than Haiku 4.5 for prompts up to 100,000 tokens, the band where most requests sit.

Published · on anthropic.com · 2 min read

Anthropic's Haiku 5.5 costs 90% less than Haiku 4.5 for prompts up to 100,000 tokens, the band where most requests sit.

Anthropic has released Claude Haiku 5.5, a small model it calls its cheapest, fastest and most capable of that class. The price is what changes plans. For requests up to 100,000 tokens, Haiku 5.5 costs 90% less than Haiku 4.5; above that threshold the reduction is 50%. Anthropic says 90% of requests to the older Haiku fell into the cheaper band. Input tokens are $0.10 per million, output $0.50, cache reads $0.01. The launch page is on anthropic.com, published on 7 October 2026.

Classification and extraction at a tenth of the price

Anthropic positions Haiku 5.5 for high-volume, cost-sensitive work and names summaries, compactions, database queries and classification requests among the tasks it handles reliably. Those are the shapes behind invoice reading and request triage: a document goes in, a small set of fields or a routing decision comes out, and the same call runs thousands of times a month. At $0.10 per million input tokens, reading every incoming invoice rather than a sample stops being a budget decision. Haiku 5.5 is also the first Haiku-class model with an adjustable effort setting, so cost and accuracy can be traded per task rather than per model.

Anthropic reports benchmark scores alongside the price: 1620 on GDPval-AA v2.1, a test of professional work across 44 occupations, against 735 for Haiku 4.5, and 72.4% on OSWorld 2.1, which measures how well an agent operates a computer to finish multi-step tasks. These are vendor-run evaluations, and the system card describes the method. Customer quotes on the page come from Asana, HubSpot, AlphaSense, Box, Rogo and Cognition. HubSpot reports 92.8% averaged over three runs on its CRM task suite; AlphaSense reports 0.84 against 0.76 for Haiku 4.5 on document question answering.

Two further changes sit alongside the launch. Cache reads on Sonnet 5.5 drop to $0.10 per million tokens from $0.20, which Anthropic says cuts the cost of most agentic work on that model by around 20%. Max and Team subscribers get monthly API credits: $100 for Max 5x, $200 for Max 20x, up to $500 pooled for Team.

Haiku 5.5 is available on Amazon Web Services, Google Cloud and Microsoft Azure, and as claude-haiku-5-5 on the Claude Platform. Anthropic states that Sonnet 5.5 and Opus 5.5 remain the better choice for complex agentic coding, and that Haiku 5.5 suits narrowly scoped tasks that were previously cost-prohibitive.

Source: anthropic.com — BARGO’s commentary on the linked source.

Event date:

Which task costs your team the most hours every week? Tell us, and we will tell you whether AI can take it and what it would cost.

← all posts · RSS