Claude Opus 4.6

published

Anthropic's most capable model

Frontier reasoning model with 1M context window. Leads on SWE-bench Verified and agentic coding tasks. Extended thinking for complex multi-step problems.

Provider anthropic Type llm Access api Params undisclosed Context 1M License proprietary

Benchmarks (4)

Benchmark	Score	Reported By	Source
SWE-bench Verified	80.8%	vendor	source ↗
MMLU	90.8%	vendor	source ↗
Aider Polyglot	89.4%	vendor	source ↗
HumanEval	92.0%	vendor	source ↗

Why It Matters

First model to break 80% on SWE-bench Verified in standard runs. The 1M context window enables full-codebase reasoning without chunking. Extended thinking mode produces step-by-step reasoning traces.

Known Limitations

Proprietary, API-only access. High output token cost. Extended thinking increases latency significantly. No open weights.

Provider anthropic

Released 2026-02-24

Training cutoff 2025-04

Created March 22, 2026

Last reconciled Never