Anthropic · 2026-10-07 · major
Claude Haiku 5.5 — Anthropic's small model gets 1M context at $0.10 input
Claude Haiku 5.5 is Anthropic's new small model. It costs $0.10 / $0.50 per million tokens for prompts up to 100K, about 90% less than Haiku 4.5, and scores 72.4% on OSWorld 2.1 against 15.7% for Haiku 4.5.
Anthropic's cheapest Claude is now a 1M-context model that can use a computer and a browser.
Key specs
| Price (input) | $0.10 / 1M tokens |
|---|---|
| Osworld 2.1 | 72.4% |
Quick facts
| Maker | Anthropic |
|---|---|
| Model ID | claude-haiku-5-5 |
| Context window | 1M tokens |
| Max output | 128K tokens |
| Knowledge cutoff | Jun 2026 |
| Default effort | medium |
| Availability | Claude API, Amazon Bedrock, Google Cloud, Microsoft Foundry, claude.ai |
Benchmarks
| Claude Haiku 5.5 | 72.4% | |
|---|---|---|
| Claude Haiku 4.5 | 15.7% | |
| GPT-6 Luna | 48.9% | |
| Claude Sonnet 5.5 | 83.9% |
| Claude Haiku 5.5 | 39.2% | |
|---|---|---|
| Claude Haiku 4.5 | 0% | |
| GPT-6 Luna | 16.4% | |
| Claude Sonnet 5.5 | 70.6% |
Pricing
| Input (≤100K prompt) | $0.10 / 1M tokens |
|---|---|
| Output (≤100K prompt) | $0.50 / 1M tokens |
| Input (>100K prompt) | $0.50 / 1M tokens |
| Output (>100K prompt) | $2.50 / 1M tokens |
| Cache read (≤100K) | $0.01 / 1M tokens |
| Batch API · input and output | 50% off |
What is it?
Claude Haiku 5.5 is the new small model in Anthropic's Claude 5.5 family, released on October 7, 2026. Anthropic calls it the cheapest, fastest and most capable small model it has released, and aims it at high-volume work such as classification, routing, extraction, summaries, live customer support and subagent tasks.
How does it work?
Adaptive thinking is on by default and is steered with the effort parameter, which defaults to medium. The context window grows to 1M tokens and max output to 128K, up from 200K and 64K on Haiku 4.5. The model also gains the browser use tool and the new computer-use toolset, and runs safety classifiers that can return a refusal. Price depends on prompt length: requests up to 100K tokens cost five times less than longer ones.
Why does it matter?
For prompts under 100K tokens the price falls about 90% compared with Haiku 4.5, while agent scores jump: 72.4% on OSWorld 2.1 against 15.7%, and 39.2% on Terminal-Bench 4.0 against 0.0%. On Anthropic's table Haiku 5.5 also beats OpenAI's GPT-6 Luna on every shared test. Teams running many cheap agent calls get a much stronger model for less money.
Who is it for?
teams running high-volume classification, routing and subagent workloads
Frequently asked questions
- How much does Claude Haiku 5.5 cost?
- Claude Haiku 5.5 costs $0.10 per million input tokens and $0.50 per million output tokens for prompts up to 100,000 tokens. Longer prompts cost $0.50 input and $2.50 output. Cache reads are $0.01 per million on short prompts, and Batch API requests take 50% off. Anthropic says that is about 90% cheaper than Claude Haiku 4.5 for short requests.
- How does Claude Haiku 5.5 compare to GPT-6 Luna?
- In Anthropic's launch table Claude Haiku 5.5 leads OpenAI's GPT-6 Luna on every shared benchmark: 72.4% vs 48.9% on OSWorld 2.1, 39.2% vs 16.4% on Terminal-Bench 4.0, 46.4% vs 42.4% on FrontierCode 1.1, and 1620 vs 1437 Elo on GDPval-AA v2.1. These are Anthropic's numbers; independent tests have not yet been published.
- What breaks when moving from Claude Haiku 4.5 to Claude Haiku 5.5?
- Several Claude Haiku 4.5 patterns return errors on Claude Haiku 5.5: manual thinking with budget_tokens, non-default temperature, top_p or top_k, assistant prefill, and the old computer_20250124 tool on the Claude API and Google Cloud. The same text also counts as about 30% more tokens, and Priority Tier is not supported. Anthropic publishes a migration guide.
- How long will Claude Haiku 5.5 stay available?
- Anthropic lists Claude Haiku 5.5 as active and commits to keeping it available until at least October 7, 2027. The model ID claude-haiku-5-5 is fixed, with no date suffix and no separate alias, so code that pins it will not silently move to a newer snapshot.
Try it
Model ID `claude-haiku-5-5` on the Claude API, Amazon Bedrock, Google Cloud and Microsoft Foundry