Claude Opus 4.8

Anthropic's upgraded flagship model, better at coding and agent work and more honest about its own progress, at the same price as Opus 4.7.

About 4x less likely to overlook flaws in code than Opus 4.7, and asks clarifying questions instead of guessing on unclear tasks
Better at controlling a browser on its own and at long legal-document work, two areas Anthropic tested directly
Same price as Opus 4.7: $5 per million input tokens and $25 per million output tokens
Fast mode dropped to about $10 input and $50 output per million tokens, roughly 3x cheaper than before
Available on claude.ai, via the API as claude-opus-4-8, and on Google Vertex AI, AWS, and Microsoft

Long, multi-step coding and agent work where the model runs on its own for a while.

You need a model that catches its own mistakes and asks before guessing on unclear tasks.

Developers and teams running Claude for code, research, and document-heavy work.

Staying informed is great. Staying ahead is better.

Pick a topic and leave with something you created.

Get new module drops and weekly AI strategy.