Claude Haiku 5.5 Cuts Small Model Prices by 75 Percent and Adds an Effort Dial

Anthropic released Claude Haiku 5.5 on 7 October 2026, and the headline is price. It costs $0.10 per million input tokens and $0.50 per million output tokens for prompts up to 100,000 tokens, against $1.00 and $5.00 for Haiku 4.5. Anthropic puts the average saving at about 75 percent.

It is also the first Haiku-class model with an adjustable effort setting. Adaptive thinking is on by default, so the model decides how much to think, and you trade quality against speed and cost with effort rather than switching thinking off. The context window grows to 1M tokens, with up to 128K tokens of output, up from 200K and 64K.

The benchmark jump is large, though these are Anthropic's own numbers. On OSWorld 2.1 (offline subset) it scores 72.4 percent against 15.7 percent for Haiku 4.5, with Sonnet 5.5 at 83.9 percent. On Terminal-Bench 4.0 it reaches 39.2 percent, while Sonnet 5.5 sits at 70.6 percent. So it closes much of the gap on computer use, but not on hard coding work.

Anthropic is clear about where to use it: summaries, compaction, classification, database queries, live customer support, browser use, and as a subagent beside Opus 5.5 or Sonnet 5.5. For complex agentic coding it still points to the larger models.

Two other changes landed alongside. Cache reads on Sonnet 5.5 were halved to $0.10 per million tokens, which Anthropic estimates cuts Sonnet 5.5 cost on most agentic work by about 20 percent. Max and Team subscribers also get a monthly API credit, from $100 on Max 5x up to $500 pooled on Team.

Our read is simple. At these prices, the routing, tagging and summarising steps inside a product no longer need a budget conversation. We build tools like Ralph, our AI project assistant, so cheap and fast small models matter to us more than another leaderboard score.

Recount your tokens before you switch, because the same text counts as roughly 30 percent more tokens than on Haiku 4.5. The model is available now on the Claude Platform, AWS, Google Cloud and Microsoft Azure as claude-haiku-5-5.