Anthropic's Claude Haiku 5.5 cuts small-model prices by up to 90%

On October 7, 2026, Anthropic released Claude Haiku 5.5, the newest version of Haiku, its family of small, fast models. Anthropic’s developer documentation describes it as built for high-volume work such as classification, routing and extraction.

For prompts up to 100,000 tokens, Haiku 5.5 costs $0.10 per million input tokens and $0.50 per million output tokens. Haiku 4.5 cost $1 and $5. Prompts longer than 100,000 tokens cost $0.50 and $2.50. Anthropic says that makes Haiku 5.5 about 90% cheaper than Haiku 4.5 for shorter requests and 50% cheaper for longer ones, and that on average it costs around 75% less to run.

The model can read up to 1 million tokens at once and write up to 128,000 tokens in one reply. Haiku 4.5 allowed 200,000 and 64,000.

There is one more change. Haiku 5.5 uses a newer tokenizer, and Anthropic’s documentation says the same text counts as about 30% more tokens than on Haiku 4.5, with the exact increase depending on the content.

Why it matters

AI companies charge by the token, a small chunk of text that is often a piece of a word. A tokenizer is the tool that cuts text into those chunks, and two tokenizers can cut the same sentence into different numbers of pieces. So a lower price per token and a higher token count push the bill in opposite directions, and the real cost depends on both.

The context window, the amount of text a model can hold in view at one time, also grew fivefold here. Our How AI Reads course explains tokens, tokenizers and context windows step by step.