Skip to main content
Claude Haiku 5.5 is Anthropic’s small, fast model from the Claude 5.5 family, released October 7, 2026, for high-volume, latency-sensitive work such as classification, extraction, routing and sub-agent tasks. It takes text and image input, outputs text, and carries a 1M token context window with up to 128k output tokens. It uses adaptive thinking steered by effort, which defaults to medium. Leave temperature, top_p and top_k unset; non-default values return a 400 error. Pricing is tiered by prompt length: when a request’s prompt (input plus cached tokens) is over 100,000 tokens, every token in that request bills at the higher tier. Both tiers are listed on the pricing page. Try it in the browser: Claude Haiku 5.5 API playground.

Model

Supported Endpoints

All three API formats are supported:
Send LLM requests to https://direct-api.unifically.com, not https://api.unifically.com. The main host sits behind Cloudflare, and Cloudflare’s proxy timeout cuts most LLM requests short before the model has finished writing. When that happens you receive a truncated reply and are still charged for every token generated up to the cut-off. direct-api.unifically.com reaches the same API without the Cloudflare proxy, so long responses complete in full.

Example

Volume discounts are available for almost all models. Reach out at contact@unifically.com to discuss custom pricing.