Direct Anthropic language and reasoning models.
13 integrated models · Official documentation ↗
Anthropic's current smallest and fastest Claude model, released October 2026 for high-volume, latency-sensitive work such as classification, extraction, routing and subagent tasks. The first Haiku with an effort control, plus a 1M-token context.
Anthropic's current Sonnet model, released September 2026 as a faster, lower-cost complement to Claude Opus 5.5. Balances speed and intelligence with a 1M-token context, adaptive thinking and a between_tools setting that skips up-front thinking.
Anthropic's current Opus model and its recommended starting point for most workloads as of October 2026, first of the Claude 5.5 family. Built for long-running agentic coding and knowledge work, with always-on adaptive thinking and a 1M-token context.
Anthropic's most capable generally available model as of October 2026, succeeding Claude Fable 5. A Mythos-class tier above Opus for demanding reasoning and long-horizon agentic work, with always-on adaptive thinking and a 1M-token context.
Opus model from July 2026 that Anthropic said came close to Claude Fable 5 at half the price. Now legacy, succeeded by Claude Opus 5.5 in September 2026; 1M-token context, 128K output and adaptive thinking.
Sonnet model from June 2026 that Anthropic called its most agentic Sonnet yet, performing close to Opus 4.8 at a lower price. Now legacy, succeeded by Claude Sonnet 5.5 in September 2026.
The first Fable model, a Mythos-class tier Anthropic placed above Opus for long, complex agentic tasks. Now legacy, superseded by Claude Fable 5.1 in September 2026; adaptive thinking is always on, with a 1M-token context and 128K output.
Legacy Opus model released May 2026 and since superseded by Claude Opus 5 and then Claude Opus 5.5. Its release focused on more reliable agentic work and more candid flagging of flaws in its own code, with a 1M-token context.
Legacy Opus model from April 2026, superseded by Opus 4.8 and the Claude 5-series Opus models. It introduced the current Claude tokenizer, image input up to 2,576 pixels on the long edge, and the xhigh effort level.
Legacy mid-tier Claude model from February 2026, superseded by Claude Sonnet 5 and then Claude Sonnet 5.5. Offers a 1M-token context and 128K output, with adaptive thinking replacing manual extended-thinking budgets.
Legacy Opus model from February 2026, several releases behind the current Claude Opus 5.5. Offers a 1M-token context and 128K output with adaptive thinking; the older manual extended-thinking mode is deprecated on this model.
Legacy Opus model from November 2025 (pinned snapshot claude-opus-4-5-20251101), superseded by later Opus releases up to Claude Opus 5.5. Has a 200K-token context, 64K output, manual extended thinking and support for the effort parameter.
Previous-generation Haiku from October 2025 (pinned snapshot claude-haiku-4-5-20251001), superseded by Claude Haiku 5.5. A small, fast model with a 200K-token context, 64K output and manual extended thinking.
