Blog
2 min read

Claude Haiku 5.5 Released: $0.10 per Million Input Tokens and a 1M Context

Anthropic launched Claude Haiku 5.5 on 7 October 2026: $0.10/$0.50 per million tokens for prompts up to 100K tokens, $0.50/$2.50 above that, a 1M-token context window, 128k output and adaptive thinking. What's new, what breaks if you're migrating from Haiku 4.5, and where it fits.

Anthropic launched Claude Haiku 5.5 (claude-haiku-5-5) on 7 October 2026, completing the Claude 5.5 family. It's built for high-volume, latency-sensitive work: classification, routing, extraction and subagents.

Pricing is tiered by prompt length

Per million tokens Prompts ≤ 100K tokens Prompts > 100K tokens
Input $0.10 $0.50
Output $0.50 $2.50
Cache read $0.01 $0.05
Batch input / output $0.05 / $0.25 $0.25 / $1.25

For comparison, Haiku 4.5 is $1 / $5. Each request is priced on its own prompt length — including cached tokens — so a request over the threshold pays the higher rate even if most of it is a cache hit. Keep prompts under 100K tokens where you can. (What Are Tokens in AI?)

What's new

  • 1M-token context window (up from 200K) and 128k output (up from 64k).
  • Adaptive thinking with the effort parameter, on by default.
  • Browser use tool support.
  • Available on the Claude API, Bedrock, Claude Platform on AWS, Google Cloud and Microsoft Foundry.

Breaking changes from Haiku 4.5

  • budget_tokens (manual extended thinking) returns an error — use adaptive thinking and effort.
  • Non-default temperature, top_p and top_k return an error — remove them.
  • Assistant prefill returns an error — end messages with a user turn.
  • Responses can start with thinking blocks — select content by type, not content[0].
  • The newer tokenizer counts the same text as roughly 30% more tokens, so recheck max_tokens and cost estimates.

(Structured Output From LLMs)

In Claude Code and elsewhere

Claude Code 2.1.293 made Haiku 5.5 the default Haiku model on the Anthropic API — used for background tasks, /goal evaluation and the haiku alias. GitHub Copilot also added it for paid plans.

Also announced the same day

Where it fits

At these prices, Haiku 5.5 makes "AI on every row" features practical: tagging every support ticket, summarising every document, routing every request. Pair it with the Batch API for another 50% off non-urgent work. (Claude Message Batches API)

Source: Claude Platform release notes · What's new in Haiku 5.5

EasySpawn gives your app a server-side backend where high-volume AI jobs can run against your database, with Claude Code to build and migrate them. See how it works or join the waitlist.

Related: Opus vs Sonnet vs Haiku · How to Build an AI App · Stop Bots Running Up Your AI Bill · Claude API Pricing Explained

Keep reading