Claude Opus 4.6 is out and filled with new things.
From Anthropic : “Opus 4.6 plans more carefully, sustains agentic tasks for longer, operates reliably in massive codebases, and catches its own mistakes.”
- 1M token context window
- 128k output tokens
- Agent teams in Claude Code
- Context compaction for agents
- Adaptive thinking
- Effort controls
-
1 on major benchmarks
- Beats GPT-5.2
- Claude in PowerPoint
- Claude in Excel upgrades
- Lowest over-refusal rate
- Same pricing
Time to check it out, let the tokens fly