Updated daily at 05:00 UTC
Company
StepFun
Shanghai AI company that makes the Step series of language, vision, audio, image and video models, several of them open-weight.
- Country
- China
- Founded
- 2023
- Headquarters
- Shanghai
- Type
- AI lab
- Open weights
- Mixed
- Website
- https://www.stepfun.com
GOT-OCR
OCR document
- GOT-OCR2.0Sep 2024CurrentOpen weights
Step
Language
- Step-1—Retired
- Step-2-mini—Retired
- Step 5 PreviewSep 2026Current
- Step 3.7 FlashMay 2026CurrentOpen weights
- Step 3.5 FlashJan 2026LegacyOpen weights
- Step3Jul 2025LegacyOpen weights
- Step-2Jul 2024Retired
Step ASR
Speech-to-text
- Step ASR—Legacy
- StepAudio 3 ASR—Current
- StepAudio 2.5 ASRApr 2026Legacy
Step Image
Image generation
- Step-1X-Medium—Retired
- Step-2X-Large—Deprecated
- NextStep-1Aug 2025LegacyOpen weights
- Step1X-EditApr 2025LegacyOpen weights
Step TTS
Text-to-speech
- Step TTS mini—Legacy
- StepAudio 3 TTS—Current
- StepAudio 2.5 TTSApr 2026Legacy
- Step TTS 2Mar 2026Legacy
Step Vision
Vision-language
- Step-1V—Retired
- Step-1o Turbo Vision—Legacy
- Step-1o Vision—Retired
- Step3-VL-10BJan 2026CurrentOpen weights
Step-Audio
Speech-to-speech
- StepAudio 2.5—Legacy
- StepAudio 3 (preview)—Current
- Step-Audio-R1Nov 2025LegacyOpen weights
- Step-Audio 2 miniAug 2025LegacyOpen weights
- Step-Audio-ChatFeb 2025LegacyOpen weights
Step-Video
Video generation
- Step-Video-TI2VMar 2025CurrentOpen weights
- Step-Video-T2VFeb 2025LegacyOpen weights