GET

/v1/schemas/fal/video/nvidia/cosmos-3-super/image-to-video

Accept:

kind inputactivity videohash 04b57d7e…extracted Aug 11, 2026version currentsource upstream spec ↗

field	recorded source
schema	Source linked above. Source trace `7299898c7aa098ef303fbf6d6128cf5f32947c0840ae91927f107cc13fb2d5bd` { "fetchedAt": 1786424549, "extractorVersion": "1", "verifiedAt": null }

Request body

16 properties · 2 required · 1 $defs

property	type	description	constraints
image_url*	string	URL of the conditioning first-frame image for the video.
prompt*	string	Text prompt describing the motion and scene of the video to generate.	length ≥ 1 · length ≤ 4096
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 …	default: true
agentic_max_iterations	integer	Maximum agentic prompt stages when agentic generation is enabled.	1 ≤ n ≤ 3 · default: 2
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.	default: false
num_inference_steps	integer	Number of denoising steps. More steps yield higher quality but take longer.	1 ≤ n ≤ 50 · default: 28
image_size	ImageSize \| enum	The size of the generated video. The request is clamped and snapped to the nearest supported NVIDIA tier (256p/480p/720p) and aspect ratio.	default: {"height":480,"width":832}
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.	length ≤ 2048 · default: "The video captures a series of …
guidance_scale	number	Classifier-free guidance scale. Higher values increase prompt adherence at the cost of diversity.	0 ≤ n ≤ 20 · default: 6
agentic_samples_per_iteration	integer	Candidate videos to generate and judge per agentic iteration. The best candidate advances to the next rewrite stage.	1 ≤ n ≤ 3 · default: 2
agentic_early_stop	boolean	Stop the agentic loop early when the critic score clears the strict quality threshold.	default: true
frames_per_second	integer	Frames per second of the output video.	4 ≤ n ≤ 60 · default: 24
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.	default: false
seed	integer	The same seed and prompt given to the same model version will produce the same video every time.	nullable
enable_safety_checker	boolean	Enable content moderation for the input prompt and image. Disabling it requires account authorization; unauthorized requests are always checked.	default: true
num_frames	integer	Number of frames to generate. More frames yield a longer video.	5 ≤ n ≤ 189 · default: 189

Types

1 $defs

ImageSizeobject

height	integer	The height of the generated image.	0 < n ≤ 14142 · default: 512
width	integer	The width of the generated image.	0 < n ≤ 14142 · default: 512

Validate a payload

before you spend tokens

shell

$ curl -X POST https://modelschemas.com/v1/validate \
    -d '{"provider":"fal","endpointId":"nvidia/cosmos-3-super/image-to-video","payload":{…}}'
→ {"valid": false, "errors": [{"path": "#", "keyword": "required", …}]}