DataLearner logo
ST

Step 3.5 Flash

Chat model

Step 3.5 Flash

Release date: 2026-02-02Updated: 2026-07-17 23:18:13.6883,233
Live demoGitHubHugging FaceCompare
Parameters
196B
Context length
256K
Chinese support
Supported
Reasoning ability

Step 3.5 Flash is a chat model from StepFunAI, released on 2026-02-02. It accepts text input and returns text output. The cataloged parameter count is 196B, with 11B active parameters per inference. The recorded context window is 256K, and the recorded maximum output is 16K. Cataloged capabilities include Reasoning model and Multilingual. The checkpoint is listed under the Apache 2.0 license, with the license link included in the model references. Use the linked references to confirm current access, licensing, and provider-specific limits.

Data sourced primarily from official releases (GitHub, Hugging Face, papers), then benchmark leaderboards, then third-party evaluators. Learn about our data methodology

Step 3.5 Flash

Model basics

Reasoning traces
Supported
Thinking modes
Standard Mode
Context length
256K tokens
Max output length
16K tokens
Model type
Chat model
Modality (in / out)
Text → Text
Release date
2026-02-02
Model file size
399GB
MoE architecture
Yes
Total params / Active params
196B / 11B
Knowledge cutoff
No data
Step 3.5 Flash

Open source & experience

Code license
Apache 2.0
Weights license
Apache 2.0- Commercial use permitted
GitHub repo
https://github.com/stepfun-ai/Step-3.5-Flash
Hugging Face
https://huggingface.co/stepfun-ai/Step-3.5-Flash
Live demo
No live demo
Step 3.5 Flash

Official resources

Paper
Paper
DataLearnerAI blog
DataLearnerAI blog
Step 3.5 Flash

API details

API speed
5/5
No public API pricing yet.
Step 3.5 Flash

Benchmark Results

Step 3.5 Flash currently shows benchmark results led by AIME2025 (6 / 107, score 99.80), LiveCodeBench (13 / 123, score 86.40), τ²-Bench (5 / 43, score 88.20). This page also consolidates core specs, context limits, and API pricing so you can evaluate the model from benchmark results and deployment constraints together.

Thinking
Tool usage

General Knowledge

2 evaluations
Benchmark / mode
Score
Rank/total
ARC-AGI
56.50
41 / 67
ARC-AGI
53.50
44 / 67

Coding and Software Engineer

2 evaluations
Benchmark / mode
Score
Rank/total
LiveCodeBench
86.40
13 / 123
SWE-bench Verified
74.40
41 / 111

Math and Reasoning

4 evaluations
Benchmark / mode
Score
Rank/total
AIME2025
99.80
6 / 107
AIME2025
97.30
18 / 107
IMO-AnswerBench
86.70
7 / 21
IMO-AnswerBench
85.40
9 / 21

Agent Level Benchmark

1 evaluations
Benchmark / mode
Score
Rank/total
τ²-Bench
88.20
5 / 43

AI Agent - Information Search

1 evaluations
Benchmark / mode
Score
Rank/total
BrowseComp
69
28 / 52

AI Agent - Tool Usage

1 evaluations
Benchmark / mode
Score
Rank/total

Claw-style Agent Evaluation

2 evaluations
Benchmark / mode
Score
Rank/total
Pinch Bench
Thinking ModeTools
85.30
15 / 37
Claw Bench
Thinking ModeTools
84.90
16 / 29
View benchmark analysisCompare with other models

Compare with other models

No curated comparisons for this model yet.

Want a custom combination? Open the compare tool

Step 3.5 Flash

Publisher

Step 3.5 Flash

Model Overview

Step 3.5 Flash is a chat model from StepFunAI, released on 2026-02-02.

It accepts text input and produces text output. Its cataloged capabilities include Reasoning model and Multilingual. The cataloged parameter count is 196B, with 11B active parameters per inference. The recorded context window is 256K, and the recorded maximum output is 16K.

The checkpoint is listed under the Apache 2.0 license, with the license link included in the model references. The evaluation section contains 13 cataloged benchmark results with their recorded modes and scores. The page links 4 release, model-card, repository, or provider references for checking the underlying claims. Specifications, availability, and prices can change; undisclosed values are intentionally left unstated.

Step 3.5 Flash

FAQ

What is Step 3.5 Flash?

Step 3.5 Flash is a chat model from StepFunAI, released on 2026-02-02. It accepts text input and returns text output. The cataloged parameter count is 196B, with 11B active parameters per inference. The recorded context window is 256K, and the recorded maximum output is 16K. Cataloged capabilities include Reasoning model and Multilingual. The checkpoint is listed under the Apache 2.0 license, with the license link included in the model references. Use the linked references to confirm current access, licensing, and provider-specific limits.

What input and output modalities does Step 3.5 Flash support?

The current model record lists text as input and text as output.

What are the main recorded specifications for Step 3.5 Flash?

The cataloged parameter count is 196B, with 11B active parameters per inference. The recorded context window is 256K, and the recorded maximum output is 16K. Fields without a source-backed value remain undisclosed.

Why is no API price shown for Step 3.5 Flash?

Open-weight checkpoint; no current first-party hosted unit price could be verified for the exact checkpoint.

Are benchmark results available for Step 3.5 Flash?

The evaluation section contains 13 cataloged benchmark results with their recorded modes and scores. Compare only results that use the same benchmark version and evaluation mode.

Is Step 3.5 Flash open source?

The checkpoint is listed under the Apache 2.0 license, with the license link included in the model references. Review the linked license text before commercial or derivative use.

DataLearner on WeChat

Follow DataLearner on WeChat for AI model updates and research notes.

DataLearner WeChat QR code