← All models
Cosmos3 is a collection of Omnimodal world models capable of generating dynamic, high-quality video, image, audio, and action commands from combinations of text, image, video, and action trajectory inputs.
Example
Details
- Accepts
- image
Pricing
- Price
- 6.25 cr / second
Prices in credits (1 credit = $0.01).
Data schema
Input
| Field | Type | Description |
|---|---|---|
| seed | — | The same seed and prompt given to the same model version will produce the same video every time. |
| promptrequired | string | Text prompt describing the motion and scene of the video to generate. |
| image_urlrequired | string | URL of the conditioning first-frame image for the video. |
| sync_mode | boolean | If `True`, the video is returned as a data URI and the output data won't be available in the request history. |
| image_size | — | The size of the generated video. The request is clamped and snapped to the nearest supported NVIDIA tier (256p/480p/720p) and aspect ratio. |
| num_frames | integer | Number of frames to generate. More frames yield a longer video. |
| guidance_scale | number | Classifier-free guidance scale. Higher values increase prompt adherence at the cost of diversity. |
| negative_prompt | string | Content to steer the generation away from (artifacts, unwanted motion). Defaults to NVIDIA's recommended i2v negative prompt; pass an empty string to disable. |
| frames_per_second | integer | Frames per second of the output video. |
| agentic_early_stop | boolean | Stop the agentic loop early when the critic score clears the strict quality threshold. |
| num_inference_steps | integer | Number of denoising steps. More steps yield higher quality but take longer. |
| enable_safety_checker | boolean | Enable content moderation for the input prompt and image. Disabling it requires account authorization; unauthorized requests are always checked. |
| agentic_max_iterations | integer | Maximum agentic prompt stages when agentic generation is enabled. |
| enable_prompt_expansion | boolean | If true, the Cosmos3-Nano Reasoner (a VLM that sees the first frame) rewrites the prompt into the dense caption Cosmos3 was trained on. The app starts a local Reasoner by default, or uses COSMOS_PROMPT_UPSAMPLER_BASE_URL when configured. Falls back to the raw prompt if expansion fails. |
| enable_agentic_generation | boolean | Enable the iterative Cosmos agentic loop: prompt upsampling, candidate video generation, VLM critique of sampled frames, and prompt rewrite. Each candidate is a full render, so this is substantially slower and costlier than a single generation. |
| agentic_samples_per_iteration | integer | Candidate videos to generate and judge per agentic iteration. The best candidate advances to the next rewrite stage. |
Output
| Field | Type | Description |
|---|---|---|
| seed | integer | The seed used for generation. |
| video | — | The generated video. |