# Media Generation

Generate images and videos, remove captions and watermarks.

### `generate_image`

Generate an AI image from a text prompt. Available on every plan. Costs 1 studio credit per image on most models (GPT Image 2.5 is 0.35, Nano Banana Pro is 2); `fast=true` costs 3x the base, minimum 3 credits.

**Example prompts**

"Generate a thumbnail image of a mysterious dark forest with glowing eyes, 16:9 aspect ratio"

"Generate a fast image of an astronaut riding a horse on Mars"

| Parameter | Type | Default | Description |
| --- | --- | --- | --- |
| prompt | string | required | Description of the image to generate |
| model | string | nano-banana-2 | Model: `nano-banana-2` (default), `nano-banana-pro` (supports up to 8 reference images), `seedream-4.5-edit` (requires reference_image_url), `seedream-5.0-lite` (optional reference), `gpt-image-2.5` (cheapest). Nano Banana 2 and the Seedream models cost 1 credit per image, Nano Banana Pro 2, `gpt-image-2.5` 0.35; `fast=true` is 3x the base, minimum 3. |
| aspect_ratio | string | 16:9 | Aspect ratio (e.g. 16:9, 9:16, 1:1) |
| reference_image_url | string | null | Optional reference image URL |
| reference_image_urls | array | null | Optional list of reference image URLs (up to 8 on nano-banana-pro; also multi-ref on nano-banana-2, seedream-5.0-lite, gpt-image-2.5). Takes precedence over reference_image_url. |
| fast | boolean | false | Fast mode — bypasses the standard ladder for sub-30s outputs. **Costs 3 credits per image instead of 1.** Defaults to `false`. |

**Fast mode:** set `fast=true` to bypass the standard provider ladder and route the request through a low-latency lane. Costs 3 credits per image instead of 1. Same models, same prompt; sub-30s outputs in the typical case. Reference images still work.

### `generate_video`

Generate an AI video from a text prompt. Available on every plan. Cost depends on model, resolution and duration, from 3 credits for a short Grok clip up to several hundred for Sora 2 Pro or 4K Seedance.

**Example prompt**

"Generate a 4-second vertical video of a cat walking through a neon-lit city at night using sora-2"

| Parameter | Type | Default | Description |
| --- | --- | --- | --- |
| prompt | string | required | Description of the video to generate |
| model | string | sora-2 | Model: `sora-2`, `kling-2.6`, `veo3-fast`, `grok-image-to-video` |
| seconds | string | 4 | Duration for sora-2 (4, 8, or 12 seconds) |
| size | string | null | Resolution (e.g. 720x1280, 1280x720) |
| duration | string | 5 | Duration for kling model (5 or 10 seconds) |
| sound | boolean | false | Include audio (kling only) |
| aspect_ratio | string | 9:16 | Aspect ratio (e.g. 9:16, 16:9) |
| input_reference_url | string | null | Reference image URL |
| resolution | string | null | Output resolution: veo3 720p/1080p/4k; grok 480p/720p/1080p; seedance 480p/720p (plus 1080p/4k on seedance-2; seedance-2-5 is 480p/720p only) |

### `get_media_job_status`

Check the status of a media generation job. Pass the job ID returned by `generate_image`, `generate_video`, or `generate_tts`; returns the job state and, when finished, the hosted result URL.

| Parameter | Type | Default | Description |
| --- | --- | --- | --- |
| job_id | string | required | Job ID returned by generate_image, generate_video, or generate_tts |

### `remove_captions`

Remove captions or watermarks from a video. Costs 12 credits per minute (0.2 per second, rounded up). Requires Professional or Ultimate plan.

| Parameter | Type | Default | Description |
| --- | --- | --- | --- |
| video_url | string | required | URL of the video to remove captions from |