Skip to content

StepFun · Chat

Step 3.7 Flash

StepFun 🇨🇳ChatStep familyOpen weights. One of its sources says the weights are publicly available.
From Empiriolabs's listing:
StepFun multimodal reasoning model with image and video input, tool calling, adjustable reasoning effort, and 256K context.
llminfo IQ
120.2
Max context
262K
Listings
36

Evidence and sources

International API storefront (platform.stepfun.ai) is Singapore-based; principal operating entity 上海阶跃星辰智能科技股份有限公司 (Shanghai).

🤗 stepfun-ai/Step-3.7-Flashllminfo IQ 120.2, ranked among chat models. From 41 benchmarks. Plenty of data behind it.How llminfo IQ is computedCoding IQ 105.6, 95% interval 90–121. From 5 benchmark families across 3 sources. A fair amount of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.How Coding IQ is computedPoor value (47.0). Costs more per task than its benchmark results predict. From 12 benchmarks with measured task costs. Plenty of data behind it.How Value for money worksYap score: tokens used per coding task vs. models with the same resultsChatty (57 ±9). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.Median 28,040 output tokens per task at default effort.How the Yap score worksReleased 23 May 2026Maker HQ: CN (source)
All rankings: #73 on OpenRouter this month #174 on OpenRouter today #197 on OpenRouter this week #515 on NanoGPT recentlySources: StepFun API data, EmpirioLabs data, Models.dev data, Hugging FacePrimary data from StepFun API from EmpirioLabs: primary mode, description from Models.dev: weights availability from Hugging Face: release date

Advantages and trade-offs

Compared with 10 chat models, picked mainly by popularity and IQ
We only make a claim when there's enough data on models of the same type. Updated 22 Nov, 06:07 UTC.Prices come from the model's maker where they sell it, otherwise their official partner, otherwise the market median. This comparison uses 8 maker's prices; 2 market median prices.

Input & output

as stated by the model's providers
Text
Images
Video
Text

Capabilities

Supported operations and provider-stated features

VisionReasoningFunction callingStructured outputPrompt cachingWeb search

Listings

 StepFunstep-3.7-flashSources: Models.dev data, StepFun API data256K$0.20cache read $0.04$1.15——
 Nous Portalstepfun/step-3.7-flash:free · freeSource: Nous Portal Inference Gateway data262K$0.00$0.00——
 Unorouterstep-3.7-flash:free · freeSource: Models.dev data256K$0.00$0.00——
 Eden AIInference by DeepInfradeepinfra/stepfun-ai/Step-3.7-FlashSource: Models.dev data262K$0.20cache read $0.04$1.15——
 Hugging FaceInference by DeepInfrastepfun-ai/Step-3.7-FlashSource: Hugging Face Inference Providers data262K$0.20$1.15——
 Kilostepfun/step-3.7-flashSources: Kilo AI Gateway data, Models.dev data256K$0.20cache read $0.04$1.15——
 LlmgatewayInference by DeepInfrastepfun-ai/Step-3.7-FlashSource: LLMGateway dataNot published by the provider$0.20cache read $0.04$1.15——
 NanoGPTstepfun/step-3.7-flash:thinking · thinkingSources: Models.dev data, NanoGPT262K$0.20cache read $0.04$1.15——
 OpenRouterInference by Novita AIstepfun/step-3.7-flash · fp8Source: OpenRouter data docs262K$0.20cache read $0.04$1.15661,897ms
 RequestyInference by Novita AInovita/stepfun/step-3.7-flashSource: Requesty data262K$0.20cache read $0.04$1.15——
Show all 29 listings

Showing 10 of 29: the maker's own listing first, then the cheapest per 1M input tokens from each provider.

Also sold by Aihubmix, Commandcode, DeepInfra, Empiriolabs, Gmicloud, Kenari, Llmgateway Providers, Novita AI, Qiniu, Vercel, Wavespeed, Zenmux.

7 listings without a price are hidden.

Included in plans

MiniStepFunSubscription$6.99/moPlan page
PlusStepFunSubscription$9.99/moPlan page
ProStepFunSubscription$29/moPlan page
MaxStepFunSubscription$99/moPlan page

Benchmarks

Named effort levels run from lowest to highest. Select a result to see its source details.

Long context

Human preference

Design Arena EloSource: Design Arena Leaderboards data · Nov 2026

Multilingual

EKI EstonianSource: Keelemudelite mõõdupuu data docs · Aug 2026

Agentic

GDPvalSource: LLM Stats data · Reported by the model's developer. The source relaying it did not verify it. claim
GDPval-AA · Published ScoreStep 3.7 Flash official README/figure · published percentage score; not Elo · Reported by StepFun · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.

Reasoning

Toolathlon VerifiedStep 3.7 Flash official README/figure · Reported by StepFun · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.

Science

GPQA Diamond2 results
Reported by StepFun · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
Step 3.7 Flash official README/figure · avg@16 · NVFP4 · MTP disabled
Step 3.7 Flash official README/figure · avg@16 · NVFP4 + MTP

Coding

Other

Claw-Eval v1.13 related benchmarks · 3 results
General (161) · Pass³
Reported by Claw-Eval · via BenchBoard docs
Claw-Eval v1.1 · general · 161 tasks · N=3 · grader: Gemini 3 Flash
General (161) · Pass@3
Reported by Claw-Eval · via BenchBoard docs
Claw-Eval v1.1 · general · 161 tasks · N=3 · grader: Gemini 3 Flash
Metric Unspecified
Reported by StepFun · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.
Step 3.7 Flash official README/figure · source does not identify metric/subset
V* · With CIStep 3.7 Flash official README/figure · with Python · Reported by StepFun · via BenchBoard docs · Reported by the model's developer. The source relaying it did not verify it.

Data credits: Epoch AI Benchmarking Hub, CC BY 4.0 (licence) · BenchBoard by SYuan03 (github.com/SYuan03/benchboard). MIT covers its code and data organization; no results licence is inferred. Numeric relays retain the original stating-source links. · Keelemudelite mõõdupuu: Independent LLM leaderboard for Estonian. Eesti Keele Instituut. https://moodupuu.eki.ee, CC BY 4.0. Latest per-benchmark results selected with partial-run flags retained. (licence) · Data by LLM Stats · EQ-Bench (Samuel J. Paech) — cite paech2023eqbench · SnakeBench (Greg Kamradt / ARC Prize Foundation), MIT (licence) · Hugging Face official benchmark leaderboards; licence varies by benchmark

Also known as: aihubmix/step-3.7-flash, deepinfra/stepfun-ai/step-3.7-flash, empiriolabs/step-3-7-flash, gmi/stepfun-ai/step-3.7-flash, novita/step-3.7-flash, novita/stepfun/step-3.7-flash, step-3-7-flash, step-3.7-flash:free, stepfun-ai/step-3.7-flash, stepfun/step-3.7-flash, stepfun/step-3.7-flash:free, stepfun/step-3.7-flash:thinking

Like this one. Tap or hover any dotted value, label or icon to see where it comes from and how it is measured. open details when tapped or hovered. How this data works