Skip to content

An editorial team and a community building with AI

vibecoding.incJoin Discord ↗
← All articles

News · Anthropic · 3 min read

Event date: · Published:

Anthropic launches Claude Haiku 5.5 and halves Sonnet 5.5 cache-read prices

On October 7, 2026, Anthropic released Claude Haiku 5.5, a small model with a 1M token context window and prices starting at $0.10/$0.50 per million tokens, and cut Sonnet 5.5 cache reads from $0.20 to $0.10.

Archive entry: this covers an event from October 7, 2026, and we published it on November 8, 2026. The details come from sources dated at the time and may have changed since.

On October 7, 2026, Anthropic released Claude Haiku 5.5 (claude-haiku-5-5), a small model tuned for high-volume and latency-sensitive work, as described in its launch announcement. The same day, Anthropic cut Claude Sonnet 5.5 prompt cache reads from $0.20 to $0.10 per million tokens.

What Haiku 5.5 offered

  • A 1M token context window, 128k max output tokens and adaptive thinking with the effort parameter.
  • Prices of $0.10 per million input tokens and $0.50 per million output tokens for prompts up to 100K tokens, and $0.50 / $2.50 for prompts over 100K.
  • Cache reads at $0.01 / $0.05 and cache writes at $0.125 / $0.625 per million tokens in the two prompt-size tiers.
  • An average cost about 75% below Claude Haiku 4.5, according to Anthropic.
  • Terminal-Bench 4.0 at 39.2% against 0.0% for Haiku 4.5, and the OSWorld 2.1 offline subset at 72.4% against 15.7%.
  • Availability on the Claude API, Claude in Amazon Bedrock, Claude Platform on AWS, Claude Models on Google Cloud (Google managed) and Claude in Microsoft Foundry.

What to check when migrating

  • Code written for Haiku 4.5 could break: manual extended thinking (budget_tokens), temperature, top_p, top_k and assistant prefill could each return a 400 error, and so could the computer_20250124 tool.
  • Adaptive thinking was on by default, and Haiku 5.5 used a tokenizer that produced more tokens for the same text.
  • Python SDK 1.12.0 and TypeScript SDK 0.132.0 added beta classes for the browser use and computer use tools, with the SDK running the tool loop, the URL and file policies and the approval callback.
  • Claude Code v2.1.293 made Haiku 5.5 the default Haiku model on the Anthropic API.

What it meant for people who build with AI

Builders of high-volume agents and sub-agents got a much cheaper small model with a 1M context. Sonnet 5.5 users got cheaper cached prompts the same day, with cache reads at 0.05x the base input price instead of 0.1x and cache writes and other prices unchanged. Further details are in the Claude Platform release notes.

Sources

Written with AI assistance, strictly from the sources listed above. Before it went live, automated checks verified its dates, sources and every language edition.

Anthropic launches Claude Haiku 5.5 and halves Sonnet 5.5 cache-read prices | vibecoding.inc