DeepSeek: DeepSeek V4 Flash wins on more metrics (5 of 9), but the right pick depends on what you optimize for — see the breakdown below.
| Metric | Anthropic: Claude Sonnet 5 | DeepSeek: DeepSeek V4 Flash |
|---|---|---|
| Intelligence Index | 53.4✓ | 40.3 |
| Coding Index | 71.5✓ | 56.2 |
| GPQA Diamond | 91%✓ | 89% |
| Design Arena Elo | 1330✓ | 1268 |
| Speed (tokens/sec) | 73 | 90✓ |
| Latency | 2.5s | 515ms✓ |
| Input price /M | $2.00 | $0.090✓ |
| Output price /M | $10.00 | $0.180✓ |
| Context window | 1M | 1.0M✓ |
| Capabilities | ReasoningToolsJSONVision | ReasoningToolsJSON |