Claude Haiku 5.5

Anthropic released Claude Haiku 5.5, a new fast, low-cost model, following an earlier promise. Its predecessor, Haiku 4.5, arrived almost a year ago priced at $1 per million input tokens and $5 per million output tokens, which was relatively expensive at the time and 10x the price of OpenAI's GPT-6 Luna, released last month.

Haiku 5.5 exactly matches GPT-6 Luna's price of $0.10/$0.50 up to 100,000 tokens. Past that threshold the price rises 5x to $0.50/$2.50, while Luna's increase at 272,000 tokens goes only to $0.20/$0.75. Haiku 5.5 also uses a new, less generous tokenizer: the same long prompt consumes roughly 1.25x as many tokens as with Haiku 4.5, which amounts to a hidden price increase. For workloads fitting within 100,000 tokens, Haiku matches Luna's price and reports higher benchmark scores; above that, Luna looks much better value.

The model does not allow reasoning to be disabled and defaults to medium. Using the llm-anthropic plugin, a low-effort pelican-on-a-bicycle SVG cost 0.0936 cents and took 7 seconds, while a max-effort version took 5 minutes 9 seconds and cost 3.3826 cents. Reasoning traces showed awareness of the benchmark. The most recent llm-anthropic release no longer requires a new plugin version for every new model.

Anthropic also halved the price of cache reads for Sonnet 5.5, and will roll out a monthly API credit for Max and Team subscribers on the Claude Platform: $100 for Max 5x users, $200 for Max 20x, and up to $500 for Team subscribers, pooled across their users. Credits are claimed via Settings, then Billing, by selecting the API organization. They do not roll over. Auto-reload can be disabled, after which API requests stop when the balance runs out.

Why it matters: developers get a cheaper fast model at parity with GPT-6 Luna only up to 100,000 tokens, with the tokenizer raising effective cost, while subscribers can apply monthly credits to API usage without rollover or billing surprises.