← all releases
StepFun AI·Feb 1, 2026·v1.0·open source

Step 3.5 Flash

Step 3.5 Flash is an open-source foundation model engineered for advanced reasoning and agentic capabilities with exceptional efficiency. Built on a sparse Mixture of Experts (MoE) architecture, it selectively activates only 11B of its 196B parameters per token, achieving a generation throughput of 100–300 tokens per second. This design allows it to rival the reasoning depth of top-tier proprietary models while maintaining the agility required for real-time interaction.

LLMLLMsOpen Source
visit website docs ↗
overview

Step 3.5 Flash is an open-source foundation model engineered for advanced reasoning and agentic capabilities with exceptional efficiency. Built on a sparse Mixture of Experts (MoE) architecture, it selectively activates only 11B of its 196B parameters per token, achieving a generation throughput of 100–300 tokens per second. This design allows it to rival the reasoning depth of top-tier proprietary models while maintaining the agility required for real-time interaction.

key features
  • 01Deep reasoning at speed with 3-way Multi-Token Prediction (MTP-3)
  • 02Robust engine for coding and agentic tasks with scalable reinforcement learning framework
  • 03Efficient long-context processing with 256K context window using Sliding Window Attention (SWA)
  • 04Accessible local deployment on high-end consumer hardware
related productsbrowse all →
Protenixopen sourceProtenix is an open-source, trainable PyTorch implementation of AlphaFold 3, designed for high-accuracy biomolecular structure prediction. It aims to advance accessible and extensible research tools for the computational biology community. ([github.com](https://github.com/bytedance/Protenix?utm_source=openai))Step-Audio-R1open sourceStep-Audio-R1 is an advanced audio language model developed by StepFun AI, designed to enhance audio reasoning capabilities by grounding its reasoning in acoustic features. It introduces Modality-Grounded Reasoning Distillation (MGRD), an iterative training framework that shifts the model's reasoning from textual abstractions to acoustic properties, effectively addressing the 'inverted scaling' problem where performance degrades with longer reasoning. This model has demonstrated superior performance across various audio understanding and reasoning benchmarks, surpassing models like Gemini 2.5 Pro and achieving results comparable to Gemini 3 Pro.Z-Imageopen sourceZ-Image is an efficient image generation foundation model developed by Tongyi-MAI, designed to produce high-quality, diverse, and stylistically versatile images. It serves as a robust base for creators, researchers, and developers seeking advanced image generation capabilities.