StepFunStep 3.7 Flash
Best configuration ranks #194 of 610 on the BenchLeader Index at 54.6 ±4.7. Last measured 29 May 2026. Released 29 May 2026.
- Blended price
- $0.416/M
- $0.185 in · $1.11 out
- Output speed
- 86 tok/s
- First answer
- 26 s
- first token 1.90 s
- Context
- 256k
- Overall index55
- Coding44
- Agents & tools65
- Maths51
- Knowledge46
- Instruction following66
- Multimodal60
- Long context63
- Composite54
Benchmark results
One column per reasoning effort. Rank is among every configuration of every model on that benchmark. Hover a score for the run it came from.
Reasoning
| Benchmark | default | Source |
|---|---|---|
| GPQA Diamond (AA)not in index | 80.9%#181 | Artificial Analysis |
| Humanity's Last Exam (AA)not in index | 21.4%#155 | Artificial Analysis |
Coding
| Benchmark | default | Source |
|---|---|---|
| SciCode | 40.0%#106 | SciCode |
| SciCode (AA)not in index | 43.9%#97 | Artificial Analysis |
Agents & tools
| Benchmark | default | Source |
|---|---|---|
| Terminal-Bench Hard | 35.6%#72 | Artificial Analysis |
| τ²-Bench Telecom (AA)not in index | 98.5%#4 | Artificial Analysis |
Maths
| Benchmark | default | Source |
|---|---|---|
| AIME 2026 | 95.0%#18 | MathArena |
| HMMT February 2026 | 87.9%#17 | MathArena |
| MathArena Apex | 14.6%#17 | MathArena |
Knowledge
| Benchmark | default | Source |
|---|---|---|
| AA-Omniscience | -37.3#244 | Artificial Analysis |
Instruction following
| Benchmark | default | Source |
|---|---|---|
| IFBench | 67.3%#87 | Artificial Analysis |
Multimodal
| Benchmark | default | Source |
|---|---|---|
| MMMU-Pro | 75.3%#75 | Artificial Analysis |
Long context
| Benchmark | default | Source |
|---|---|---|
| AA-LCR | 73.7%#125 | Artificial Analysis |
Composite
| Benchmark | default | Source |
|---|---|---|
| AA Intelligence Index | 19.5#179 | Artificial Analysis |
Where to run it
Every provider serving this model through OpenRouter, with throughput and first-token latency measured on live traffic over the last 30 minutes and each provider’s own price. Purple marks the best in each column.
| Provider | Speed | First token | Input $/M | Output $/M | Context | Quantisation |
|---|---|---|---|---|---|---|
| DeepInfra | 135 tok/s | 0.20 s | $0.160 | $0.920 | 262k | – |
| NovitaAI | 89 tok/s | 1.90 s | $0.200 | $1.15 | 262k | fp8 |
| StepFun | 77 tok/s | 3.25 s | $0.200 | $1.15 | 256k | fp8 |
Price history
Listed price per 1M tokens over time, as recorded by OpenRouter for the provider with the longest history.
What a task costs
Estimates from list price, output speed and time to first answer for the best configuration. “With caching” assumes three-quarters of the input is served from the prompt cache at $0.037 per 1M. Reasoning tokens are not modelled.
| Workload | Tokens in / out | Cost | With caching | Time |
|---|---|---|---|---|
| Chat reply | 400 / 300 | $0.0004 | $0.0004 | 29.6 s |
| Summarise a 30-page report | 12,000 / 600 | $0.0029 | $0.0016 | 33.1 s |
| Code edit | 6,000 / 1,500 | $0.0028 | $0.0021 | 43.5 s |
| Agentic coding session | 60,000 / 4,000 | $0.015 | $0.0089 | 1.2 min |
| Structured extraction | 2,000 / 200 | $0.0006 | $0.0004 | 28.4 s |
See also
Data as of 9 Sept 2026. Compare with another model.