Claude Haiku 5.5, explained: Anthropic’s cheapest small model gets a 1M-token window
Anthropic released Claude Haiku 5.5 on Oct. 7, 2026, and called it "the cheapest, fastest, and most capable small model we've ever released" in its launch post. The model starts at $0.10 per million input tokens and can read up to 1 million tokens in one request, compared with 200,000 tokens for Claude Haiku 4.5, according to Anthropic's Haiku 5.5 documentation.
What is Claude Haiku 5.5, in one sentence?
Claude Haiku 5.5 is the smallest and fastest model in Anthropic's current Claude lineup, built for high-volume, cost-sensitive work such as summaries, database queries, classification, live customer support and acting as a helper, or subagent, for the larger Claude Opus 5.5 and Claude Sonnet 5.5.
How much does Claude Haiku 5.5 cost compared with Haiku 4.5?
Claude Haiku 5.5 has two prices that depend on prompt length, while Haiku 4.5 has one. The Haiku 4.5 documentation compares the two:
Claude Haiku 5.5 vs. Claude Haiku 4.5, as of Nov. 6, 2026
| Claude Haiku 5.5 | Claude Haiku 4.5 | |
|---|---|---|
| Input price, per million tokens | $0.10 (prompts up to 100K tokens); $0.50 (over 100K) | $1 |
| Output price, per million tokens | $0.50 (up to 100K); $2.50 (over 100K) | $5 |
| Context window | 1M tokens | 200K tokens |
| Max output | 128K tokens | 64K tokens |
| Thinking | Adaptive, with effort setting | Extended (manual) |
| Reliable knowledge cutoff | June 2026 | February 2025 |
| Status | Active (latest) | Legacy, still available |
Anthropic says Haiku 5.5 costs about 75% less than Haiku 4.5 on average: 90% less for requests up to 100,000 tokens and 50% less for longer ones. Haiku 5.5 uses a newer tokenizer, so the same text counts as about 30% more tokens than on Haiku 4.5, the documentation says. Anthropic says its savings estimate already accounts for that.
How did Anthropic measure Haiku 5.5?
Anthropic published scores on eight benchmarks covering knowledge work, computer use, coding and visual reasoning. It compared Haiku 5.5 with Haiku 4.5, OpenAI's GPT-6 Luna and its own Sonnet 5.5. These are the company's own results. The launch post also quotes early-testing results from customers such as Asana, HubSpot and Box.
What did Anthropic find?
Claude Haiku 5.5 scored far above Haiku 4.5 on every test where both were reported, according to the launch post:
- Computer use: 72.4% on OSWorld 2.1 (offline subset), against 15.7% for Haiku 4.5 and 83.9% for Sonnet 5.5.
- Agentic coding: 39.2% on Terminal-Bench 4.0, against 0.0% for Haiku 4.5 and 70.6% for Sonnet 5.5.
- Hard questions: 45.9% on Humanity's Last Exam without tools, against 10.2% for Haiku 4.5.
Anthropic says Sonnet 5.5 and Opus 5.5 are still the better choice for complex agentic coding, and that Haiku 5.5 fits narrower tasks such as summarizing or subagent work.
Haiku 5.5 is also the first Haiku model with an adjustable effort setting, which lets developers trade cost for more reasoning. The default is medium, according to the documentation.
On safety, Anthropic reports improvements across almost all of its alignment evaluations compared with Haiku 4.5. Its cybersecurity safeguards are stricter than Haiku 4.5's and still block penetration testing. Organizations doing broader security or biology work can apply to Anthropic's Cyber Verification Program or Life Sciences Verification Program.
What doesn't Anthropic say?
The launch post does not say whether Claude Haiku 5.5 is available to free Claude.ai users or in the Claude apps. Anthropic has also not announced a shutdown date for Haiku 4.5. The documentation lists Haiku 4.5 as a legacy model with retirement "not sooner than October 15, 2026," and says it "is still available."
What should developers do?
Developers can call claude-haiku-5-5 on the Claude API, Amazon Bedrock, Google Cloud and Microsoft Foundry. Before switching, check two changes in the documentation: Haiku 5.5 uses adaptive thinking, controlled by the effort setting, instead of Haiku 4.5's manual extended thinking, and it does not accept the temperature, top_p or top_k settings. Anthropic's documentation lists a retirement date for Haiku 5.5 "not sooner than October 7, 2027."
In the same launch, Anthropic cut the cache-read price of Sonnet 5.5 by half, to $0.10 per million tokens.
FAQ
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, and $0.50 and $2.50 for longer prompts, according to Anthropic's documentation.
Is Claude Haiku 5.5 better than Haiku 4.5?
On Anthropic's own benchmarks, yes, by wide margins. For example, it scored 72.4% on the OSWorld 2.1 computer-use test, against 15.7% for Haiku 4.5.
What is the context window of Claude Haiku 5.5?
1 million tokens, with up to 128,000 tokens of output per request. Claude Haiku 4.5 has a 200,000-token window.
Is Claude Haiku 4.5 being retired?
Anthropic lists Haiku 4.5 as a legacy model that is still available, with retirement not sooner than Oct. 15, 2026. As of Nov. 6, 2026, no shutdown date had been announced.
Sources
Cite this
Claude Haiku 5.5, explained: Anthropic’s cheapest small model gets a 1M-token window, New Mind Museum, November 6, 2026. https://newmindmuseum.com/models/2026/11/claude-haiku-5-5-explained-price-context/
New Mind Museum. (2026, November 6). Claude Haiku 5.5, explained: Anthropic’s cheapest small model gets a 1M-token window. https://newmindmuseum.com/models/2026/11/claude-haiku-5-5-explained-price-context/
“Claude Haiku 5.5, explained: Anthropic’s cheapest small model gets a 1M-token window.” New Mind Museum, 6 Nov. 2026, https://newmindmuseum.com/models/2026/11/claude-haiku-5-5-explained-price-context/.
@misc{new-mind-museum-2026-claude-haiku-5-5-explained-anthropics-ch,
author = {{New Mind Museum}},
title = {Claude Haiku 5.5, explained: Anthropic’s cheapest small model gets a 1M-token window},
year = {2026},
url = {https://newmindmuseum.com/models/2026/11/claude-haiku-5-5-explained-price-context/}
}Free to quote and cite with a link to this page.