Claude Sonnet 4.6

Claude Sonnet 4.6 is Anthropic's sweet spot between speed and intelligence, and the recommended default for most production workloads. For the first time, a Sonnet model surpasses the previous generation's Opus on coding evaluations — in developer preference tests, 70% pick Sonnet 4.6 over Sonnet 4.5 and 59% pick it over Opus 4.5 — delivering near-Opus capability at roughly 1/5 of Opus pricing.

Core Capabilities 1M context window: Opus-class long-context capability — fits an entire book or a large codebase. Near-Opus coding: 79.6% on SWE-bench Verified, within roughly 1 point of Opus 4.6. Computer use: 94% accuracy on insurance-industry benchmarks — well suited to browser and desktop automation. Adaptive Thinking: Recommended — the model decides reasoning depth automatically. Highly competitive pricing: ~$3/M input and ~$15/M output — about 1/5 the cost of Opus 4.6. Technical Specifications Model ID: claude-sonnet-4-6 Context window: 1,000,000 tokens Max output: 64,000 tokens (up to 300K on the Batch API via beta header) Multimodal: Text + image input, text output Computer use: Compatible with the newer computer-use-2025-11-24 beta header Typical Use Cases Production coding agents — handles 90%+ of everyday coding tasks Long-context document analysis, code audits, and compliance review Customer service automation, content generation, email drafting, and other business workflows Browser and desktop automation via computer use Core generation model for RAG systems Recommendations Unless you have a clear reason to pick something else, make Sonnet 4.6 your default. Escalate to Opus 4.7 or 4.6 only when: the task is genuinely hard reasoning (proofs, complex multi-step planning, deep refactors across many files), you're seeing bad Sonnet output that better prompting can't fix, or you're running long-chain agentic work where early errors cascade. Drop down to Haiku 4.5 for simple classification, extraction, or dialogue routing.

Available On Claude API, Claude Platform on AWS, Amazon Bedrock, Google Vertex AI, and Microsoft Foundry.