---
name: Cosmos-Predict1-5B
slug: cosmos-predict1-5b
type: Multimodal
providers:
  - nvidia
description: 'Video model for prompt-guided generation, editing, and motion workflows'
tags:
  - open-source
  - multimodal
is_open_source: true
max_output: 4096
pricing:
  - image: 0
    input: 0
    output: 0
    provider: nvidia
release_date: '2025-03-18'
capabilities:
  audio: false
  vision: true
  tool_use: false
  citations: false
  reasoning: false
---
Cosmos-Predict1-5B is a Multimodal model.
Video model for prompt-guided generation, editing, and motion workflows