| Feature | Claude Sonnet 4.6 | Claude Haiku 4.5 |
|---|---|---|
| Developer | Anthropic | Anthropic |
| Released | 2026-02-17 | 2025-10-15 |
| Input Price | $3/1M tokens | $1/1M tokens |
| Output Price | $15/1M tokens | $5/1M tokens |
| Context Window | 1M tokens (beta) | 200K tokens |
| Output Tokens | 128K | 64K |
| Thinking Mode | Adaptive thinking + Extended thinking | Extended thinking — no adaptive thinking |
| Multimodal | Yes — text and images | Yes — text and images |
| API Available | Yes — API, Bedrock, Vertex AI, Foundry | Yes — API, Bedrock, Vertex AI, Foundry |
| Best For | Complex coding, large codebase understanding, long-form writing, multi-step reasoning, production coding agents, enterprise RAG, instruction-following workflows | High-volume automation, real-time chat interfaces, agentic routing and subtask execution, batch processing, classification, summarization, low-latency tool use |
| SWE-Bench Verified | 79.6% ✓ | ~65% |
| Coding Score (Relative) | 94/100 ✓ | 82/100 |
| Tool Use Score (Relative) | 96/100 ✓ | 88/100 |
| Context Window | 1M tokens ✓ | 200K tokens |
| Output Tokens | 128K ✓ | 64K |
| Input Price | $3/1M | $1/1M ✓ |
| Output Price | $15/1M | $5/1M ✓ |
| Speed | Fast | Near-instant ✓ |
| Knowledge Cutoff | August 2025 ✓ | February 2025 |
| Adaptive Thinking | Yes ✓ | No |
Claude Sonnet 4.6 vs Haiku 4.5: Which Claude Model Should You Use?
By Joe ~ July 2, 2026