STEP3-VL-10B

Open weights Active Multimodal Dense

10B open vision-language model with a PaCoRe parallel reasoning mode.

Released
Jan 14, 2026
Context window
Not disclosed
Max output
Not disclosed
Parameters
10B
Input price
Not disclosed
Output price
Not disclosed

Prices are first-party list prices in USD at the standard tier. Last verified Oct 30, 2026.

Pricing

No first-party API. This is an open-weights release you can run yourself.

Specifications

Developer
StepFun
Model type
Multimodal
Architecture
Dense
Parameters
10B
Input
Text Image
Output
Text
Knowledge cutoff
Not disclosed
License
Apache License 2.0
API access
No first-party API
Capabilities
Reasoning Small LM Vision