Claude Sonnet 4.6
Claude Sonnet 4.6 is a Claude 4 model from Anthropic, released in 2025-09. It costs $3 / $15 per 1M, has a 1M-token context window, and is best for coding, agents, rag. Last verified 2026-04-19.
Spec sheet
Pricing
- Input
- $3 / 1M
- Output
- $15 / 1M
- Cached input
- $0.3 / 1M
- Batch discount
- 50%
- Free tier
- No
Context & speed
- Context window
- 1M tokens
- Max output
- 64k tokens
- Throughput
- ~85 tok/s
- Time to first token
- ~600 ms
- Speed tier
- balanced
Capabilities
- Tool use
- Yes
- Structured output
- Yes
- Prompt caching
- Yes
- Extended thinking
- Yes
- Vision input
- Yes
- Audio in / out
- No
- Fine-tuning
- No
Deployment
- Open weights
- No
- On-prem
- No
- HIPAA eligible
- Yes
- Zero retention
- Yes
- Regions
- us, eu
Estimated monthly cost
Assumes typical token shape: 2k input, 600 output per call. Prompt caching is excluded from these figures.
When to use Claude Sonnet 4.6
Sweet spot
- coding
- agents
- rag
- long documents
- general production
Known trade-offs
- slightly below Opus on hardest reasoning
Best use cases
Best LLM for Autonomous agents
Best-in-class tool use + long-horizon planning (τ-bench leader)
Best LLM for Coding assistant
Current benchmark leader on real-world coding + agents
Best LLM for Long documents (100k+ tokens)
1M context with excellent retrieval + prompt caching
Best LLM for RAG
1M context + prompt caching (up to 90% cost cut on repeat prompts)
Best LLM for Vision
Best for UI/document understanding and diagrams
Best LLM for Content writing
Nearly as good as Opus, 5× cheaper
Works with
Compare Claude Sonnet 4.6 to other models
FAQ — Claude Sonnet 4.6
How much does Claude Sonnet 4.6 cost?
Claude Sonnet 4.6 costs $3 / $15 per 1M tokens on the Anthropic API. Cached input reads cost $0.3 per 1M, cutting the input bill by roughly 90% on repeat system prompts. The batch API offers a 50% discount for async workloads.
What is the context window of Claude Sonnet 4.6?
Claude Sonnet 4.6 has a 1M-token context window with up to 64k tokens of output. That's enough for entire codebases, long transcripts, or multi-document RAG.
Does Claude Sonnet 4.6 have a free tier?
No — Free via Claude.ai web chat; API requires paid credits.
Is Claude Sonnet 4.6 HIPAA / EU / on-prem friendly?
Claude Sonnet 4.6 is HIPAA-eligible, available in EU regions, and is API-only. Zero data retention is available for enterprise customers.
What is Claude Sonnet 4.6 best for?
Claude Sonnet 4.6 is best for coding, agents, rag, long documents. Trade-offs to be aware of: slightly below Opus on hardest reasoning.
Which tools and SDKs work with Claude Sonnet 4.6?
Claude Sonnet 4.6 integrates with Anthropic SDK, Claude Code, Cursor, AWS Bedrock, Google Vertex AI, Vercel AI SDK, LangChain, LlamaIndex, OpenRouter. Most major AI frameworks support it either natively or through OpenAI-compatible endpoints.