Anthropic's Claude Haiku 5.5 cuts the cost of its fastest model by about 75%
Claude Haiku 5.5 is Anthropic's new small model for fast, high-volume jobs. It comes with a 1 million token context window, a much lower price and one catch: the same text now counts as more tokens.
By ScoviAI Team
Key points
- Released October 7, 2026 as Anthropic's fastest and most capable small model
- API price for prompts up to 100K tokens: $0.10 per million input and $0.50 per million output tokens
- 1 million token context window and up to 128K tokens of output
- The same text produces about 30% more tokens than on Haiku 4.5
Anthropic has released Claude Haiku 5.5, the newest version of its smallest and fastest Claude model. Anthropic says it costs roughly 75% less to run than Haiku 4.5 while doing more.
Haiku is the model companies use when they need answers quickly and in large numbers: sorting support tickets, summarizing documents, routing requests, or running as a helper inside a bigger AI agent.
What does Claude Haiku 5.5 cost?
On the Claude API, the price depends on how long your prompt is:
| Prompt size | Input | Output |
|---|---|---|
| Up to 100K tokens | $0.10 per 1M tokens | $0.50 per 1M tokens |
| Over 100K tokens | $0.50 per 1M tokens | $2.50 per 1M tokens |
Most everyday requests fall in the first row. Very long prompts, such as a whole book or a big codebase, cost five times more per token.
The tokenizer catch
Anthropic changed how Haiku 5.5 splits text into tokens. The same text produces about 30% more tokens than on Haiku 4.5. The per-token price is still far lower, but if your app has budgets or limits set in tokens, you'll need to recount them. We explain what tokens are, and why this matters for your bill, in AI tokens, explained.
What's new besides the price
- A 1 million token context window and up to 128K tokens of output in one answer.
- An effort setting, the first on a Haiku model, so developers can trade speed for more careful answers.
- Adaptive thinking by default, which lets the model decide how much to think before it replies.
Anthropic recommends Haiku 5.5 for summarizing, classifying, routing and helper tasks, and for jobs where speed matters, such as live support and browsing the web for a user. For hard coding work, Anthropic says its larger Sonnet 5.5 and Opus 5.5 models remain the better choice.
Where you can use it
Haiku 5.5 is available on the Claude API under the model ID claude-haiku-5-5, on Amazon Bedrock, Google Cloud and Microsoft Foundry, and in Claude.ai and Claude Code.
What it means for you
- Developers running high-volume jobs on Haiku 4.5 can likely cut costs a lot, after recounting tokens.
- Claude app users may notice faster replies on quick tasks; the bigger models are still there for hard ones.
If you're thinking about a paid Claude plan, our Claude Pro price guide explains what each plan costs.
Sources
- Introducing Claude Haiku 5.5 (Anthropic)
- What's new in Claude Haiku 5.5 (Claude Platform Docs)
- Pricing (Claude Platform Docs)
ScoviAI Team is the ScoviAI newsroom. We write every article ourselves from the original announcements and link our sources. Spotted a mistake? Tell us and we'll fix it.
ADVERTISEMENT
