Claude Opus 4.8

Anthropic's flagship Opus upgrade, just 41 days after Opus 4.7 — a "modest but tangible" improvement. AA Intelligence Index v4.1: 56. Headline gains are in honesty (more likely to acknowledge when it lacks information, less likely to make unsupported claims) and code reliability (~4× less likely than 4.7 to let flaws in its generated code go unremarked). API name claude-opus-4-8.

Shipped with a "dynamic workflow" capability (Claude orchestrating multiple subagents at once), a user-facing effort control (how much reasoning to spend per response), and a fast mode (~2.5× speed, ~3× cheaper). Available via claude.ai, Claude Code, the API, Amazon Bedrock, Google Cloud Vertex AI, and Microsoft Foundry.