StepFunStep 3.7 Flash
Step 3.7 Flash is a StepFun open-weights reasoning model, released 29 May 2026. Its best configuration ranks #343 of 785 on the BenchLeader Index at 51.5 ±5.0, in the upper half. The ± is the point: 328 other configurations score within that range, so they and this one cannot be told apart on quality alone — price and speed are what separate them. It scores highest in instruction following (67) and lowest in coding (40). At $0.438 per million tokens blended it is cheaper than most ranked models. Output speed of 179 tokens per second puts it in the fastest quarter, with a first answer in 13.6 s. Last measured 29 May 2026.
- Blended price
- $0.438/M
- $0.200 in · $1.15 out
- Output speed
- 179 tok/s
- measured by Artificial Analysis
- First answer
- 14 s
- first token 3.72 s
- Context
- 256k
- Full answer
- 16 s
- median, reasoning included
- Cost per run
- –
- Released
- 29 May 2026
- Overall index52
- Reasoning45
- Coding40
- Agents & tools48
- Maths51
- Knowledge45
- Instruction following67
- Multimodal58
- Long context62
- Composite51
Versions
StepFun has shipped 3 models under this name. Each is ranked on its own results; a newer version often has fewer results so far, which holds its index nearer the average until more arrive.
| Model | Released | Index | Rank |
|---|---|---|---|
| Step 3.7 Flashthis page | 29 May 2026 | 51.5 | #343 |
| Step 3.5 Flash | 2 Apr 2026 | 53.7 | #275 |
| Step 3.5 Flash 0202 | 2 Feb 2026 | 50.7 | #365 |
How its index has moved
9 Sept 2026 to 18 Oct 2026The index is recomputed from scratch every day, so a line moves when a new benchmark result lands, when a publisher revises a score, or when the models it is normalised against change. Early movement usually means the score is still settling.
Benchmark results
See all LLM benchmarks. 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 | not stated | Source |
|---|---|---|
| GPQA Diamond (AA)not in index | 80.9%#186 | Artificial Analysis |
| Humanity's Last Exam (AA)not in index | 21.4%#209 | Artificial Analysis |
| CritPt | 2.3%#198 | Artificial Analysis |
Coding
| Benchmark | not stated | Source |
|---|---|---|
| SciCode | 40.0%#162 | SciCode |
| SciCode (AA)not in index | 43.9%#152 | Artificial Analysis |
| ALE-Benchnot in index | 694.1#86 | Epoch AI Benchmarking Hub |
| Terminal-Bench 2.1 (AA)not in index | 39.3%#140 | Artificial Analysis |
Agents & tools
| Benchmark | not stated | Source |
|---|---|---|
| Terminal-Bench Hard | 35.6%#74 | Artificial Analysis |
| τ²-Bench Telecom (AA)not in index | 98.5%#4 | Artificial Analysis |
| τ³-Banking (AA) | 12.0%#134 | Artificial Analysis |
| ITBench SRE (AA)not in index | 30.3%#34 | Artificial Analysis |
| APEX-Agents (AA)not in index | 14.8%#23 | Artificial Analysis |
| EnterpriseOps-Gymnot in index | 29.1%#42 | Artificial Analysis |
Maths
| Benchmark | not stated | Source |
|---|---|---|
| AIME 2026possible contamination | 95.0%#18 | MathArena |
| HMMT February 2026possible contamination | 87.9%#17 | MathArena |
| MathArena Apexpossible contamination | 14.6%#17 | MathArena |
Knowledge
| Benchmark | not stated | Source |
|---|---|---|
| AA-Omniscience | -37.3#311 | Artificial Analysis |
| AA-Omniscience: accuracynot in index | 25.8%#224 | Artificial Analysis |
| AA-Omniscience: non-hallucinationnot in index | 15.0%#339 | Artificial Analysis |
Instruction following
| Benchmark | not stated | Source |
|---|---|---|
| IFBench | 67.3%#88 | Artificial Analysis |
Multimodal
| Benchmark | not stated | Source |
|---|---|---|
| MMMU-Pro | 75.3%#103 | Artificial Analysis |
Long context
| Benchmark | not stated | Source |
|---|---|---|
| AA-LCR | 73.7%#173 | Artificial Analysis |
Composite
| Benchmark | not stated | Source |
|---|---|---|
| AA Intelligence Index v4.3.2 | 19.5#240 | 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 |
|---|---|---|---|---|---|---|
| NovitaAI | 78 tok/s | 3.57 s | $0.200 | $1.15 | 262k | fp8 |
| StepFun | 55 tok/s | 4.59 s | $0.200 | $1.15 | 256k | fp8 |
Price history
The price above, per 1M tokens, every day it changed.
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 $0.040 per 1M. Reasoning tokens are not modelled.
| Workload | Tokens in / out | Cost | With caching | Time |
|---|---|---|---|---|
| Chat reply | 400 / 300 | $0.0004 | $0.0004 | 15.3 s |
| Summarize a 30-page report | 12,000 / 600 | $0.0031 | $0.0017 | 17.0 s |
| Code edit | 6,000 / 1,500 | $0.0029 | $0.0022 | 22.0 s |
| Agentic coding session | 60,000 / 4,000 | $0.017 | $0.0094 | 36.0 s |
| Structured extraction | 2,000 / 200 | $0.0006 | $0.0004 | 14.7 s |
See also
Questions people ask
- What are Step 3.7 Flash's benchmark scores?
- As of 24 Oct 2026, Step 3.7 Flash scores 51.5 ±5.0 on the BenchLeader Index, strongest in instruction following (67) and weakest in coding (40). That rests on 12 independent benchmark results, and every score on this page links to its publisher.
- How does Step 3.7 Flash rank?
- Step 3.7 Flash is #343 of 785 ranked configurations on 24 Oct 2026. 328 other configurations score within its margin, so the exact position matters less than the band. The leader is Claude Opus 5.5 (max effort) at 72.6.
- How much does Step 3.7 Flash cost per 1M tokens?
- As of 24 Oct 2026, Step 3.7 Flash costs $0.200 per 1M input tokens and $1.15 per 1M output tokens, $0.438 blended (three input tokens per output token). The list price last changed on 6 Oct 2026.
- When was Step 3.7 Flash released?
- StepFun released Step 3.7 Flash on 29 May 2026.
Data as of 24 Oct 2026. Compare with another model.
Cite as: BenchLeader, “Step 3.7 Flash: benchmarks, pricing, speed and rank”, data as of 24 Oct 2026, https://www.benchleader.com/models/step-3-7-flash.