Models Directory/nvidia/cosmos3-nano
nvidiaCatalog Onlyfree endpoint

cosmos3-nano

nvidia/cosmos3-nano

Generates images, videos and action predictions from text, visual inputs and spatial controls, with mandatory input screening and visual SynthID watermarking.

Context Length
262,144 tokens
Input Pricing
Not published
Output Pricing
Not published
Recent Latency
No probes

Model Architecture & Limits

Canonical IDnvidia/cosmos3-nano
Publisher / Originatornvidia
Model FamilyCosmos
ArchitectureDiffusion Transformer (DiT)
Context Window Limit262,144 tokens
Max Output Tokens4,096 tokens
Endpoint Protocolnim-custom

Capabilities & Modal Surfaces

Chat Conversation✕ No
Tool / Function Calling✕ No
Reasoning / Thinking✕ No
Structured Output (JSON)✕ No
Vision / Multimodal✓ Yes
Audio Processing✕ No
License & Restrictions
NVIDIA API Trial Terms of Service
Commercial and research use permitted under governing model and API terms.