Claude Haiku 5.5
Anthropic's smallest model, now priced at $0.10 per million input tokens with a 1M context window
Context window
1M
Input / 1M tokens
$0.10
Output / 1M tokens
$0.50
Provider
Anthropic
For prompts up to 100K tokens; above 100K it is $0.50/$2.50. Cache reads $0.01 per 1M; Batch is 50% off. Uses a newer tokenizer that produces about 30% more tokens than Haiku 4.5 for the same text, so the effective saving vs Haiku 4.5 is closer to 75% than the 90% list-price cut. · Data verified 2026-10-08
Claude Haiku 5.5 (October 7, 2026) is Anthropic's fast, low-cost model. It cuts Haiku's list price by 90% to $0.10/$0.50 per million tokens, matching OpenAI's GPT-6 Luna, while raising the context window to 1M tokens and max output to 128K. It runs adaptive thinking by default. Anthropic reports large gains over Haiku 4.5 on agentic tasks, and Artificial Analysis scores it between 29 and 43 on its Intelligence Index depending on effort setting.
Capability index
Relative estimates (0-100) to place this model against its peers, grounded in published benchmarks.
How to access it
Generally available on the Claude API (model ID claude-haiku-5-5), Amazon Bedrock (anthropic.claude-haiku-5-5), Google Cloud, Microsoft Foundry and Claude Platform on AWS.
Strengths
- ✓Lowest Claude price ever: $0.10/$0.50 per 1M tokens (prompts up to 100K)
- ✓1M context window and 128K max output
- ✓Strong agentic results for its size: 72.4% OSWorld 2.1 (Anthropic-reported)
- ✓Adaptive thinking with adjustable effort levels
Best for developers who...
When to choose it (and when not to)
Reach for Claude Haiku 5.5 when...
- →You run high volumes of simple or mid-complexity calls and cost dominates
- →You need a 1M context window at the lowest price
- →You are migrating from Haiku 4.5 and want a large cost cut
Look elsewhere if...
- ✕Your code reads content[0], sends temperature, or uses assistant prefill (all break on Haiku 5.5)
- ✕You need frontier-level reasoning on hard problems
Benchmarks
| Benchmark | Score | Notes |
|---|---|---|
| OSWorld 2.1 | 72.4% | Anthropic-reported; Haiku 4.5 15.7%, GPT-6 Luna 48.9% |
| Terminal-Bench 4.0 | 39.2% | Anthropic-reported; GPT-6 Luna 16.4% |
| GDPval-AA v2.1 | 1620 Elo | Anthropic-reported; GPT-6 Luna 1437 |
| Artificial Analysis Intelligence Index | 43 (max effort) / 34 (medium) | Third-party; GPT-6 Luna 38 |
Source: Anthropic announcement
Tools powered by Claude Haiku 5.5
Compare Claude Haiku 5.5
Compare Claude Haiku 5.5 with any other model
Build a comparison →All model comparisons →