Trust
the
Bench
Leaderboard
Models
Benchmarks
Audio
Image
Video
Labs
More
Search
⌘K
Skip to main content
StepFun
Headquarters: Shanghai, China
Official Website →
Category:
AI Lab
Chinese AI lab developing the Step family of language and audio models.
Text & multimodal models
2
Open-weight models
2
Audio models
5
Image and video models
2
First tracked release
2026-01-20
Latest tracked release
2026-09-18
Top Intelligence Index
28.5 · Step 5
Models by StepFun (2)
S
StepFun
Sep 2026
Open weights
Step 5
Not yet measured
Context
1M
Speed
—
Size
Undisclosed
S
StepFun
Jan 2026
Open weights
Step3 VL 10B
Not yet measured
Context
66k
Speed
—
Size
Undisclosed
Image and video models by StepFun (2)
Step Image Edit 2
Image editing · Elo 1008 · May 2026
Step1X-Edit
Image editing · Elo 589 · Apr 2025
Audio models by StepFun (5)
StepAudio 2.5 TTS (Aug 2026)
Text-to-speech · Elo 1215 · Aug 2026
Step TTS 2 (Mar 2026)
Text-to-speech · Elo 1136 · Feb 2026
Step Audio EditX (Mar 2026)
Text-to-speech · Elo 1100 · Feb 2026
StepAudio 2.5 ASR, StepFun
Speech-to-text · WER 4.7%
StepAudio 3 ASR, StepFun
Speech-to-text · WER 2.5%