Step 5 Preview
Open weights promised for 2026-10-15 (placeholder HF repo stepfun-ai/Step-5-Preview-BF16); update open_weights/license then. Pricing from press, not verified on StepFun's pricing page.
- Context window
- 1,000,000 tokens
- Max output
- 64,000 tokens
- Input
- text, image, video
- Output
- text
- License
- proprietary
- Pricing
- input: $1 · cached input: $0.05 · output: $2.7 (per 1M tokens (USD); reasoning tokens billed as output) source
- Verified
- 2026-09-29
How to call it
| Provider | Model id | Endpoint / URL | Docs |
|---|---|---|---|
| StepFun API | step-5-preview | — | docs |
Notable capabilities (1)
- 600B sparse MoE agent model with 1M context and video input: 600B total / 27B active, 92 layers; up to 60 images and short videos per request; reasoning effort low/medium/high. source
Timeline entry
- StepFun launches Step 5 Preview, a 600B-parameter MoE agent model with 1M context; open weights promised for Oct 15 ★★★
On Sept 20, 2026 StepFun released Step 5 Preview (`step-5-preview`), its flagship agentic model: a 600B total / 27B active sparse MoE with 92 layers, a 1M-token context and 64K max output, taking text, up to 60 images and video. It costs $1.00 input ($0.05 cached) and $2.70 output per 1M tokens…
Other StepFun models
StepAudio 3 ASR Max / StepAudio 3 TTS · StepFun Step-Audio-EditX · StepAudio 3 Realtime