StepFun: Step 3.5 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 4.5 | StepFun: Step 3.5 Flash |
|---|---|---|
| Intelligence Index | 37.1 | 37.8✓ |
| Coding Index | 33.5✓ | 31.6 |
| GPQA Diamond | 73% | 83%✓ |
| Design Arena Elo | 1242✓ | — |
| Speed (tokens/sec) | 45✓ | 44 |
| Latency | 787ms | 418ms✓ |
| Input price /M | $3.00 | $0.090✓ |
| Output price /M | $15.00 | $0.300✓ |
| Context window | 1M✓ | 262K |
| Capabilities | ReasoningToolsJSONVision | ReasoningToolsJSON |