Claude Haiku 5.5
Anthropic · Multimodal model
Claude Haiku 5.5 is Anthropic's small, fast model, released on October 7, 2026. It costs a fraction of the larger Claude models and is meant for high-volume work and as a sub-agent, not for complex coding.
Current
ReleasedOct 7, 2026
Context window1M tokens (128K max output)
Input → outputtext, image → text
API priceFrom $0.10 / input MTok, From $0.50 / output MTok ($0.10/$0.50 for prompts up to 100k tokens; $0.50/$2.50 above 100k)
Consumer accessClaude Free, Pro, Max, Team and Enterprise; also in Claude Code
Best for
High-volume, latency-sensitive tasks such as classification, extraction, routing, summarization and sub-agent work.
Overview
Claude Haiku 5.5 is Anthropic's small, fast model, released on October 7, 2026. It costs a fraction of the larger Claude models and is meant for high-volume work and as a sub-agent, not for complex coding.
Strengths and limits
- Cheapest and fastest current Claude model; about 75% cheaper on average than Haiku 4.5
- First Haiku with effort controls (low to max)
- Large jump over Haiku 4.5 on knowledge-work and computer-use evaluations
- Anthropic recommends Sonnet 5.5 or Opus 5.5 for complex agentic coding
- Price rises 5x for prompts over 100k tokens
Published benchmark scores
Terminal-Bench 4.039.2%
Source: www.anthropic.com
OSWorld 2.1 (offline subset)72.4%
Source: www.anthropic.com
Humanity's Last Exam (with tools)57.4%
Source: www.anthropic.com
GDPval-AA v2.11620
Source: www.anthropic.com
Scores are as published by the source; settings (reasoning effort, tools, attempts) differ between labs, so compare with care. How to read benchmarks →