Step3
Multimodal modelStep3-321B-A38B
Step3-321B-A38B is an AI model published by StepFunAI, released on 2025-07-31, for Multimodal model, with 321B parameters, and 64K context length, requiring about 642GB storage, with a 82.90 score on AIME2025.
Data sourced primarily from official releases (GitHub, Hugging Face, papers), then benchmark leaderboards, then third-party evaluators. Learn about our data methodology
Model basics
Open source & experience
Official resources
API details
Benchmark Results
Step3 currently shows benchmark results led by LiveCodeBench (101 / 250, score 67.10), AIME2025 (88 / 215, score 82.90), MMMU (37 / 74, score 74.20). This page also consolidates core specs, context limits, and API pricing so you can evaluate the model from benchmark results and deployment constraints together.
Multimodal Understanding
2 evaluationsModel variants & downloads
| Variant name | Version type | Quantization | Model size | HuggingFace link |
|---|---|---|---|---|
| Step3-FP8ℹ️ | Instruct | FP8 | 327.11 GB | Download link |
Publisher
Model Overview
Step3-321B-A38B is an AI model published by StepFunAI, released on 2025-07-31, for Multimodal model, with 321B parameters, and 64K context length, requiring about 642GB storage, with a 82.90 score on AIME2025.
DataLearner on WeChat
Follow DataLearner on WeChat for AI model updates and research notes.
