Claude Haiku 5.5
Anthropic released Claude Haiku 5.5, its smallest model, with a 1M token context window at $0.10 per million input tokens and $0.50 output, about 75% cheaper than Haiku 4.5 on average.
Haiku 5.5 succeeds Haiku 4.5 with a 1M token context window, 128k max output tokens, and adaptive thinking with an effort setting, which is a first for a Haiku model. Anthropic says it is its fastest model and suits summaries, classification, browser use and subagent coding work alongside Opus 5.5 and Sonnet 5.5. Anthropic also halved Sonnet 5.5 cache read prices and added a monthly API credit for Max and Team subscribers.
- Date
- Wednesday, 7 October 2026
- Lab
- Anthropic
- Kind
- model
- Access
- closed API
- Price
- $0.10 input / $0.50 output per 1M tokens up to 100K tokens; $0.50 / $2.50 above
Figures
| Measure | Value | Measured by |
|---|---|---|
| GDPval-AA v2.1 | 1620 Haiku 4.5 scored 735 | company |
| OSWorld 2.1 (offline subset) | 72.4% Haiku 4.5 scored 15.7% | company |
| Terminal-Bench 4.0 | 39.2% Haiku 4.5 scored 0.0% | company |
| Humanity's Last Exam (no tools) | 45.9% Haiku 4.5 scored 10.2% | company |
| Input price per 1M tokens | $0.10 requests up to 100K tokens; $0.50 above, per The New Stack via Techmeme | company |
| Output price per 1M tokens | $0.50 requests up to 100K tokens; $2.50 above, per The New Stack via Techmeme | company |
Available on the Claude API, Amazon Bedrock, Claude Platform on AWS, Google Cloud and Microsoft Foundry. Code written for Haiku 4.5 can break because manual budget_tokens thinking returns a 400 error.
Sources
- anthropic.com/claude-haiku-5-5
- docs.claude.com/en/release-notes/overview
- tradingview.com/news/seekingalpha:cee58c21e094b:0-anthropic-unveils-low-cost-claude-haiku-
This record was checked against its sources on 7 October 2026. How we check