Claude Opus 4.8 launched with 1M token context and improved adaptive thinking

We've launched Claude Opus 4.8 (claude-opus-4-8), our most capable generally available model. Claude Opus 4.8 supports a 1M token context window by default on the Claude API, Amazon Bedrock, Google Cloud, and Microsoft Foundry, 128k max output tokens, and the same set of tools and platform features as Claude Opus 4.7.

Key improvements and new features:

• Mid-conversation system messages: Send role: "system" messages after a user turn in the messages array, preserving prompt cache hits when instructions change during a long-running session.
• Stop details for refusals are now publicly documented, returning a category (cyber, bio, or null) and a human-readable explanation.
• Effort parameter defaults to high across all surfaces, including Claude Code and the Messages API.
• Prompt caching minimum cacheable prompt length is 1,024 tokens, lower than Claude Opus 4.7.
• Adaptive thinking triggers reasoning only when a turn needs it, reducing wasted thinking tokens.
• Supports high-resolution image input (up to 2576 pixels on the long edge).
• Task budgets, advisor tool, and computer use now supported.

See What's new in Claude Opus 4.8 for capability improvements, new features, and migration guidance.