Anthropic
Claude Sonnet 4.6
The default workhorse Claude — 60% of Opus quality questions at 60% of the price.
API onlytextimagefileREL 2026-02-17
Compare this model →Verified 2026-06-12
Overview
Claude Sonnet 4.6 is Anthropic's mid-tier: strong everyday coding and reasoning with a 1M-token context window at $3/$15 per 1M tokens. It is the model most Claude Code users run day-to-day, escalating to Opus only for the hardest problems.
Benchmark scores
Methodology →Intelligence52
Codingn/a
Reasoningn/a
Mathn/a
Sources and verification
- Intelligence: Artificial Analysis Intelligence Index · checked 2026-06-12
- Catalogue and pricing: OpenRouter model record · checked 2026-06-12
Strengths and weaknesses
Strengths
- Excellent cost/quality balance
- 1M-token context
- Fast enough for interactive coding
Trade-offs
- A clear step below Opus on hardest tasks
- Closed weights
Reach for this model when
Daily-driver coding assistant and general assistant work.
Run something like it locally
Open-weights peers at this tier: Kimi K2.6, GLM-5.1, Qwen3.7 — all runnable via vLLM on multi-GPU rigs.
