Model comparison
Claude Sonnet 4.6 vs DeepSeek R1
Anthropic
Claude Sonnet 4.6
The pragmatic default - Claude quality without Opus pricing.
DeepSeek
DeepSeek R1
The open-weights reasoning model that reset the cost curve.
Specs
| Metric | Claude Sonnet 4.6 | DeepSeek R1 |
|---|---|---|
| Context window | 1M tokens↑ | 64K tokens |
| Input $/1M tokens | $3.00 | $0.700↑ |
| Output $/1M tokens | $15.00 | $2.50↑ |
| Modalities | Text · Image · File | Text |
| Open weights | No | Yes |
| Released | - | Jan 2025 |
Capability differences
| Capability | Claude Sonnet 4.6 | DeepSeek R1 |
|---|---|---|
| Tool use | Yes | No |
| Vision | Yes | No |
| Prompt caching | Yes | No |
| Open weights | No | Yes |
Key differences
- -Claude Sonnet 4.6 offers 1M context; DeepSeek R1 offers 64K.
- -Input pricing: Claude Sonnet 4.6 at $3.00/1M vs DeepSeek R1 at $0.700/1M.
- -DeepSeek R1 is open weights; Claude Sonnet 4.6 is proprietary.