nvidia/cosmos-3-super/image-to-video

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.

Inference

Commercial use

Schema

LLMs

Playground API Examples

1. Calling the API

2. Authentication

API Key

3. Queue

4. Files

5. Schema

About

Generate a video from a first-frame image and prompt using Cosmos3-Super via vLLM-Omni.

1. Calling the API#

Install the client#

The client provides a convenient way to interact with the model API.

npm install --save @fal-ai/client

Migrate to @fal-ai/client

The @fal-ai/serverless-client package has been deprecated in favor of @fal-ai/client. Please check the migration guide for more information.

Setup your API Key#

Set FAL_KEY as an environment variable in your runtime.

export FAL_KEY="YOUR_API_KEY"

Submit a request#

The client API handles the API submit protocol. It will handle the request status updates and return the result when the request is completed.

import { fal } from "@fal-ai/client";

const result = await fal.subscribe("nvidia/cosmos-3-super/image-to-video", {
  input: {
    prompt: "The camera slowly pushes in as the subject turns their head toward the light, hair drifting in a gentle breeze, dust motes floating through warm afternoon sun.",
    image_url: "https://storage.googleapis.com/falserverless/example_inputs/hunyuan_i2v.jpg"
  },
  logs: true,
  onQueueUpdate: (update) => {
    if (update.status === "IN_PROGRESS") {
      update.logs.map((log) => log.message).forEach(console.log);
    }
  },
});
console.log(result.data);
console.log(result.requestId);

2. Authentication#

The API uses an API Key for authentication. It is recommended you set the FAL_KEY environment variable in your runtime when possible.

API Key#

In case your app is running in an environment where you cannot set environment variables, you can set the API Key manually as a client configuration.

import { fal } from "@fal-ai/client";

fal.config({
  credentials: "YOUR_FAL_KEY"
});

Protect your API Key

When running code on the client-side (e.g. in a browser, mobile app or GUI applications), make sure to not expose your FAL_KEY. Instead, use a server-side proxy to make requests to the API. For more information, check out our server-side integration guide.

3. Queue#

Long-running requests

For long-running requests, such as training jobs or models with slower inference times, it is recommended to check the Queue status and rely on Webhooks instead of blocking while waiting for the result.

Submit a request#

The client API provides a convenient way to submit requests to the model.

import { fal } from "@fal-ai/client";

const { request_id } = await fal.queue.submit("nvidia/cosmos-3-super/image-to-video", {
  input: {
    prompt: "The camera slowly pushes in as the subject turns their head toward the light, hair drifting in a gentle breeze, dust motes floating through warm afternoon sun.",
    image_url: "https://storage.googleapis.com/falserverless/example_inputs/hunyuan_i2v.jpg"
  },
  webhookUrl: "https://optional.webhook.url/for/results",
});

Fetch request status#

You can fetch the status of a request to check if it is completed or still in progress.

import { fal } from "@fal-ai/client";

const status = await fal.queue.status("nvidia/cosmos-3-super/image-to-video", {
  requestId: "764cabcf-b745-4b3e-ae38-1200304cf45b",
  logs: true,
});

Get the result#

Once the request is completed, you can fetch the result. See the Output Schema for the expected result format.

import { fal } from "@fal-ai/client";

const result = await fal.queue.result("nvidia/cosmos-3-super/image-to-video", {
  requestId: "764cabcf-b745-4b3e-ae38-1200304cf45b"
});
console.log(result.data);
console.log(result.requestId);

4. Files#

Some attributes in the API accept file URLs as input. Whenever that's the case you can pass your own URL or a Base64 data URI.

Data URI (base64)#

You can pass a Base64 data URI as a file input. The API will handle the file decoding for you. Keep in mind that for large files, this alternative although convenient can impact the request performance.

Hosted files (URL)#

You can also pass your own URLs as long as they are publicly accessible. Be aware that some hosts might block cross-site requests, rate-limit, or consider the request as a bot.

Uploading files#

We provide a convenient file storage that allows you to upload files and use them in your requests. You can upload files using the client API and use the returned URL in your requests.

import { fal } from "@fal-ai/client";

const file = new File(["Hello, World!"], "hello.txt", { type: "text/plain" });
const url = await fal.storage.upload(file);

Auto uploads

The client will auto-upload the file for you if you pass a binary object (e.g. File, Data).

Read more about file handling in our file upload guide.

5. Schema#

Input#

prompt string* required

Text prompt describing the motion and scene of the video to generate.

image_url string* required

URL of the conditioning first-frame image for the video.

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. Default value: "The video captures a series of frames showing macroblocking artifacts, chromatic aberration, high-frequency noise, and rolling shutter distortion. It includes static with no motion, motion blur, over-saturation, shaky footage, low resolution, grainy texture, pixelated images, poorly lit areas, underexposed and overexposed scenes, poor color balance, washed out colors, choppy sequences, jerky movements, low frame rate, bit-depth compression artifacts, color banding, unnatural transitions, outdated special effects, fake elements, unconvincing visuals, poorly edited content, jump cuts, hard cut, visual noise, and flickering. It features moiré patterns, edge halos, and temporal aliasing. Furthermore, the content defies common sense, generating illogical scenarios, nonsensical entities, absurd character behaviors, and conceptual paradoxes that violate basic human reasoning and everyday reality. The video looks like a surreal or glitchy hallucination. Overall, the video is of poor quality."

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. Default value: true

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_max_iterations integer

Maximum agentic prompt stages when agentic generation is enabled. Default value: 2

agentic_samples_per_iteration integer

Candidate videos to generate and judge per agentic iteration. The best candidate advances to the next rewrite stage. Default value: 2

agentic_early_stop boolean

Stop the agentic loop early when the critic score clears the strict quality threshold. Default value: true

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.

Possible enum values: square_hd, square, portrait_4_3, portrait_16_9, landscape_4_3, landscape_16_9

Note: For custom image sizes, you can pass the width and height as an object:

"image_size": {
  "width": 1280,
  "height": 720
}

num_frames integer

Number of frames to generate. More frames yield a longer video. Default value: 189

fps integer

Frames per second of the output video. Default value: 24

num_inference_steps integer

Number of denoising steps. More steps yield higher quality but take longer. Default value: 28

guidance_scale float

Classifier-free guidance scale. Higher values increase prompt adherence at the cost of diversity. Default value: 6

seed integer

The same seed and prompt given to the same model version will produce the same video every time.

enable_safety_checker boolean

Enable content moderation for the input prompt and image. Default value: true

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.

{
  "prompt": "The camera slowly pushes in as the subject turns their head toward the light, hair drifting in a gentle breeze, dust motes floating through warm afternoon sun.",
  "image_url": "https://storage.googleapis.com/falserverless/example_inputs/hunyuan_i2v.jpg",
  "negative_prompt": "The video captures a series of frames showing macroblocking artifacts, chromatic aberration, high-frequency noise, and rolling shutter distortion. It includes static with no motion, motion blur, over-saturation, shaky footage, low resolution, grainy texture, pixelated images, poorly lit areas, underexposed and overexposed scenes, poor color balance, washed out colors, choppy sequences, jerky movements, low frame rate, bit-depth compression artifacts, color banding, unnatural transitions, outdated special effects, fake elements, unconvincing visuals, poorly edited content, jump cuts, hard cut, visual noise, and flickering. It features moiré patterns, edge halos, and temporal aliasing. Furthermore, the content defies common sense, generating illogical scenarios, nonsensical entities, absurd character behaviors, and conceptual paradoxes that violate basic human reasoning and everyday reality. The video looks like a surreal or glitchy hallucination. Overall, the video is of poor quality.",
  "enable_prompt_expansion": true,
  "agentic_max_iterations": 2,
  "agentic_samples_per_iteration": 2,
  "agentic_early_stop": true,
  "image_size": {
    "height": 480,
    "width": 832
  },
  "num_frames": 189,
  "fps": 24,
  "num_inference_steps": 28,
  "guidance_scale": 6,
  "enable_safety_checker": true
}

Output#

video VideoFile* required

The generated video.

seed integer* required

The seed used for generation.

{
  "video": {
    "content_type": "video/mp4",
    "url": "https://v3b.fal.media/files/b/0a8fc99c/cosmos3-i2v-example.mp4"
  },
  "seed": 1143
}

Other types#

ImageSize#

width integer

The width of the generated image. Default value: 512

height integer

The height of the generated image. Default value: 512

ImageFile#

url string* required

The URL where the file can be downloaded from.

content_type string

The mime type of the file.

file_name string

The name of the file. It will be auto-generated if not provided.

file_size integer

The size of the file in bytes.

width integer

The width of the image

height integer

The height of the image

VideoFile#

url string* required

The URL where the file can be downloaded from.

content_type string

The mime type of the file.

file_name string

The name of the file. It will be auto-generated if not provided.

file_size integer

The size of the file in bytes.

width integer

The width of the video

height integer

The height of the video

fps float

The FPS of the video

duration float

The duration of the video

num_frames integer

The number of frames in the video

nvidia/cosmos-3-super/image-to-video

Table of contents

1. Calling the API

2. Authentication

3. Queue

4. Files

5. Schema

About

1. Calling the API#

Install the client#

Migrate to @fal-ai/client

Setup your API Key#

Submit a request#

2. Authentication#

API Key#

Protect your API Key

3. Queue#

Long-running requests

Submit a request#

Fetch request status#

Get the result#

4. Files#

Data URI (base64)#

Hosted files (URL)#

Uploading files#

Auto uploads

5. Schema#

Input#

Output#

Other types#

ImageSize#

ImageFile#

VideoFile#

Related Models