Quality
Highest overall scores
- 1
StepFun: Step 3.7 Flash30.3 - 2StepFun: Step 3.5 Flash0
StepFun model benchmarks.
Compare StepFun models across answer quality, coding, complex tasks, speed, price, and context size.
Response speed
Fastest output in this sample
- 1StepFun: Step 3.7 Flash187.5 t/s
- 2StepFun: Step 3.5 Flash56 t/s
Input price
Lowest price per 1M tokens
- 1StepFun: Step 3.5 Flash$0.10/1M
- 2StepFun: Step 3.7 Flash$0.20/1M
Highlights
Quality
Overall capability · Higher is better
Speed
Output tokens per second · Higher is better
Input price
USD per 1M tokens · Lower is better
Quality benchmarks
Compare StepFun model capability.
Performance and price
Compare practical tradeoffs.
All StepFun models.
2 models in the current catalogue.
| Model | Links | |||||
|---|---|---|---|---|---|---|
| StepFun: Step 3.7 Flash | 30.3 | 187.5 t/s | 262.1K | $0.20 | $1.15 | |
| StepFun: Step 3.5 Flash | 0 | 56 t/s | 262.1K | $0.10 | $0.30 |
StepFun technical benchmarks.
Compare named evaluation records available for StepFun models. Missing results remain blank rather than being estimated.
| Benchmark | Area | Step 3.7 Flash | Step 3.5 Flash | Source |
|---|---|---|---|---|
| Intelligence Index | overallQuality | 30.3 | — | OpenRouter / artificial-analysis |
| Coding Index | coding | 39.6 | — | OpenRouter / artificial-analysis |
| Agentic Index | agentic | 21.5 | — | OpenRouter / artificial-analysis |
| GPQA | GPQA Diamond | 80.9 | 83.1 | OpenRouter model benchmarks |
| Humanity's Last Exam | HLE | 19.9 | 19.1 | OpenRouter model benchmarks |
| IFBench | IFBench | 67.3 | 64.6 | OpenRouter model benchmarks |
| τ²-Bench Telecom | τ²-Bench Telecom | 98.5 | 94.4 | OpenRouter model benchmarks |
| AA-LCR | AA-LCR | 63.7 | 43 | OpenRouter model benchmarks |
| GDPval-AA | GDPval-AA | 25.8 | — | OpenRouter model benchmarks |
| CritPt | CritPt | 2.3 | 2.5 | OpenRouter model benchmarks |
| SciCode | SciCode | 40 | 40.4 | OpenRouter model benchmarks |
| Terminal-Bench Hard | Terminal-Bench Hard | 35.6 | 27.3 | OpenRouter model benchmarks |
| AA-Omniscience Accuracy | AA-Omniscience Accuracy | 25.4 | 23.9 | OpenRouter model benchmarks |
| AA-Omniscience Non-Hallucination Rate | AA-Omniscience Non-Hallucination Rate | 15.6 | 14.8 | OpenRouter model benchmarks |
| MMLU-Pro | Knowledge and reasoning | — | 84.4 | OpenEvals/leaderboard-data |
| SWE-bench Verified | Coding | — | 74.4 | OpenEvals/leaderboard-data |
| Terminal-Bench | Agentic coding | — | 51 | OpenEvals/leaderboard-data |
| AIME 2026 | Math | — | 96.67 | OpenEvals/leaderboard-data |