Claude Opus 4.6
Claude Opus 4.6 adds a 1M-token context (beta), 128K output, adaptive thinking and Claude Code agent teams at unchanged $5/$25 pricing.
First Opus with a 1M-token window (beta; premium $10/$37.50 beyond 200K tokens). Adaptive thinking with four effort levels, context compaction (beta), and agent teams in Claude Code (research preview). Anthropic reports MRCR v2 8-needle at 1M tokens of 76% vs 18.5% for Sonnet 4.5, and GDPval-AA 144 Elo above GPT-5.2.
- Date
- Thursday, 5 February 2026
- Lab
- Anthropic
- Kind
- model
- Access
- closed API
- Price
- $5/$25 per M tokens in/out; above 200K tokens $10/$37.50; fast mode research preview from 2026-02-07
Figures
| Measure | Value | Measured by |
|---|---|---|
| MRCR v2 (8-needle, 1M tokens) | 76% vs 18.5% for Sonnet 4.5 | company |
| GDPval-AA | +144 Elo vs GPT-5.2 +190 Elo vs Opus 4.5; Artificial Analysis benchmark, reported by Anthropic | independent |
| Terminal-Bench 2.0 | 65.4% as cited in Anthropic's Mythos Preview comparison | company |
| SWE-bench Verified | 80.8% as cited in Anthropic's Mythos Preview comparison | company |
1M context reached general availability at standard pricing on 2026-03-13. The page also launched Claude in PowerPoint (research preview). Prefill of assistant turns was removed in this generation. Rival comparisons on the page are against GPT-5.2.
Sources
- www.anthropic.com/news/claude-opus-4-6
- www.anthropic.com/glasswing
- platform.claude.com/docs/en/release-notes/overview
This record was checked against its sources on 6 October 2026. How we check