Skip to content
RankShark

Step 3.7 Flash

Step 3.7 Flash is StepFun's latest high-efficiency multimodal Mixture-of-Experts model.

ModelOpen sourceby StepFunsince 2026
27.2

SharkScore

#268 in Models

Why it scores 27.2

Step 3.7 Flash has a SharkScore of 27.2/100, ranked #268 of 607 in Models (updated 2026-11-19).

Each part compares Step 3.7 Flash with other listings in Models. Evidence covers 87% of the score's weight.

Adoption
0 · 30%
User reviews
no data · 25%
Maintenance
17 · 20%
Trust & transparency
80 · 15%
Momentum
no data · 10%

Paying never changes this score. How ranking works · method v2026-11-18.2

RankShark review of Step 3.7 Flash

Automated evaluation by RankShark · 18 Nov 2026 · method v1.0 · How we review

We were not paid for this review.

As of 2026-11-18, Step 3.7 Flash is a multimodal model with a 262,144-token context window and token-based pricing listed by OpenRouter. [E1][E2]

Step 3.7 Flash is StepFun’s sparse Mixture-of-Experts model for text and image input; OpenRouter also lists video input. The model card describes a 198B-parameter model with about 11B parameters active per token. [E1][E2]

Best for

  • Developers building workflows that combine image understanding, reasoning, and tool use. · model card, OpenRouter
  • Teams seeking a model they can run through hosted providers or local inference frameworks. · model card

Key features

  • Multimodal input. The vendor describes native image understanding; OpenRouter lists text, image, and video input modalities. · model card, OpenRouter
  • Long context. The vendor says the model supports a 256k context window; OpenRouter lists 262,144 tokens. · model card, OpenRouter
  • Selectable reasoning effort. The vendor says users can select low, medium, or high reasoning levels. · model card
  • Local inference options. The vendor lists vLLM, SGLang, Hugging Face Transformers, and llama.cpp as supported inference backends. · model card
  • Sparse MoE architecture. The vendor describes a 198B-parameter model that activates approximately 11B parameters per token. · model card

Pros

  • The model card lists an Apache-2.0 license. · model card
  • The vendor documents several local inference backends and deployment options. · model card
  • OpenRouter lists text, image, and video as input modalities. · OpenRouter

Cons

  • The vendor’s llama.cpp instructions list 120 GB as the minimum unified memory or VRAM for that local setup. · model card
  • No free plan or free trial stated on the pages we checked.
  • Whether customer data is used for model training is not stated on the pages we checked.

Pricing

PlanPriceLimits
Input (cache miss)$0.20 / M tokensInput tokens; cache miss.
Input (cache hit)$0.04 / M tokensInput tokens; cache hit.
Output$1.15 / M tokensOutput tokens.

Checked 18 Nov 2026

Free plan
Not stated (checked 18 Nov 2026)
Free trial
Not stated (checked 18 Nov 2026)
Trains on customer data
Not stated (checked 18 Nov 2026)
Data retention
Not stated (checked 18 Nov 2026)
Data regions
Not stated (checked 18 Nov 2026)
Data processing agreement
Not stated (checked 18 Nov 2026)
AI models used
Step 3.7 Flash: a 198B-parameter sparse Mixture-of-Experts model, with a 196B language backbone and a 1.8B vision encoder; approximately 11B parameters are active per token. · model card
API
The vendor provides a StepFun API example. OpenRouter lists chat completions, completions, messages, and responses endpoints. · model card, OpenRouter
SDKs
The vendor’s Python API example uses the OpenAI client. · model card
Webhooks
Not stated (checked 18 Nov 2026)
Platforms
The vendor lists StepFun Open Platform (Global and China), OpenRouter, and NVIDIA NIM as available platforms, and describes cloud, data-center, and local deployment options. · model card
Security claims
No security certification stated (checked 18 Nov 2026)

What we measured

Context window (OpenRouter listing) · now
262,144 tokens
Price per million tokens (OpenRouter listing) · now
$0.2 in, $1.15 out
Package downloads · 30 days
9,746
How we tested · sources
  • Hourly uptime and response-time checks of the product's URL.
  • Repository, release and download data from GitHub, npm, PyPI and Hugging Face where they exist.
  • We did not sign up, pay, or use the product as a customer.

Something wrong? Report an error with a public link that shows it; corrections re-run the review.

Top alternatives to Step 3.7 Flash

See all
  • 67.1

    Qwen3.8 27B is an open-weight dense vision-language model from Qwen.

    ModelOpen source
  • 65.5

    Qwen3.6-35B-A3B is an open-weight multimodal model from Alibaba Cloud with 35 billion total parameters and 3 billion active parameters per t

    ModelOpen source
  • 64.7

    Qwen3.8 Flash is a multimodal reasoning model from Alibaba.

    ModelOpen source
  • 64.3

    Qwen3.6 27B is a dense 27-billion-parameter language model from the Qwen Team at Alibaba, released in April 2026.

    ModelOpen source