Step 3.7 Flash (High) AI Evaluation | LayerLens - Benchmark Results & Performance

Model

Step 3.7 Flash (High) by StepFun

text+image+video->text

Free credits included. No credit card required.

Step 3.7 Flash is StepFun's latest high-efficiency multimodal Mixture-of-Experts model. It pairs a 196B-parameter language backbone with a vision encoder for native image and video understanding, activating roughly 11B parameters per token. The model supports a 256K context window and exposes selectable reasoning levels, letting callers trade off speed, cost, and depth of reasoning. Designed for coding, agentic workflows, structured outputs, and long-context productivity tasks. Reasoning level is set to High.

Evaluations

across 9 benchmarks

Latency

19.7s

Context length

256k

tokens

Cost

$0.20 · $1.15

input · output per 1M tokens

Key takeaways

Key takeaways will appear here once evaluations are run.

Technical specifications

Context length: 256k tokens
Max output: —
Architecture: MoE
Modalities: text+image+video->text
Parameters: ~198B
Open weights: Yes
Licence: Apache 2.0
Regions: China
Vendor: StepFun
Provider model key: stepfun/step-3.7-flash