StepFun: Step 3.7 Flash wins on more metrics (6 of 9), but the right pick depends on what you optimize for — see the breakdown below.
| Metric | DeepSeek: DeepSeek V3.2 | StepFun: Step 3.7 Flash |
|---|---|---|
| Intelligence Index | 41.7 | 42.6✓ |
| Coding Index | 36.7 | 37.1✓ |
| GPQA Diamond | 84%✓ | 81% |
| Design Arena Elo | 1221 | 1232✓ |
| Speed (tokens/sec) | 41 | 82✓ |
| Latency | 458ms✓ | 1.5s |
| Input price /M | $0.229 | $0.200✓ |
| Output price /M | $0.343✓ | $1.15 |
| Context window | 131K | 256K✓ |
| Capabilities | ReasoningToolsJSON | ReasoningToolsJSONVision |