Claude Haiku model releases

Claude Haiku 5.5 is Anthropic's fast, low-cost small model released 2026-10-07, succeeding Claude Haiku 4.5. Anthropic calls it its cheapest and fastest small model to date, and it is the first Haiku-class model with an adjustable effort control (adaptive thinking on by default, effort defaulting to medium). Text and image input, text output; a 1,000,000-token context window and 128,000 max output tokens (up to 300,000 via the Batch API in beta); reliable knowledge cutoff June 2026. Proprietary and API-only, generally available via the Claude API as claude-haiku-5-5 and on Amazon Bedrock, Google Cloud Vertex AI, Microsoft Foundry, and the Claude Platform on AWS. Tiered pricing: $0.10 / 1M input and $0.50 / 1M output for prompts up to 100k tokens, rising to $0.50 / 1M input and $2.50 / 1M output above 100k; $0.01 / 1M cache reads and $0.125 / 1M 5-minute cache writes, with a further 50% off via batch processing. Reported benchmarks: OSWorld 2.1 72.4%, Terminal-Bench 4.0 39.2%, FrontierCode 1.1 46.4%, Humanity's Last Exam 45.9% (57.4% with tools), GDPval-AA v2.1 1620.

Anthropic's fast, low-cost Claude 4.5 model, released in October 2025 for latency-sensitive coding, tool-use, and customer-facing agent workloads. Anthropic positions it as bringing near-Sonnet capability to the Haiku tier at substantially lower cost and higher speed.