Video generation, editing and extension
Make videos with sound from a prompt or an image with Gemini Omni Flash and Grok Imagine, then edit them in words or extend them, through the API or in workspace chat.
> /v1/videos
boldrouter now makes videos. Describe a scene, or attach a picture to bring to life, and get back an MP4 with sound. Then change it in plain words ("make it look like a pencil sketch") or continue it from its last frame.
Three video models
- Gemini Omni Flash (
google/gemini-omni-1.1-flash): 3 to 10 seconds, 360p to 4K, landscape or portrait. Billed per token, about $0.10 a second at 720p. - Grok Imagine Video 1.5 (
xai/grok-imagine-video-1.5): 1 to 15 seconds, 480p to 1080p, from up to 3 reference images. $0.08, $0.14 or $0.25 a second at 480p, 720p or 1080p. - Grok Imagine Video 1.5 Lite (
xai/grok-imagine-video-1.5-lite): the same lengths and sizes for less. $0.02, $0.04 or $0.14 a second.
Grok also makes square, 4:3 and 3:2 videos.
Through the API
A video takes from about 15 seconds to a few minutes, so it runs in the background. Start it, check its status, and download the file when it's ready:
curl https://boldrouter.com/v1/videos \
-H "Authorization: Bearer $BOLD_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "xai/grok-imagine-video-1.5-lite",
"prompt": "A paper boat drifting on a calm pond at sunrise",
"resolution": "720p",
"seconds": 6
}'Then poll GET /v1/videos/{id} until status is completed, and fetch the MP4 from GET /v1/videos/{id}/content. The request and response follow OpenAI's video API, so size and input_reference work too.
Edit and extend
POST /v1/videos/{id}/remixapplies a change you describe and keeps everything else.POST /v1/videos/{id}/extendcontinues the video from its last frame by 2 to 10 seconds. You get the whole video back.
Gemini Omni edits and extends the videos it made, so nothing is uploaded again; extensions can take a video up to 40 seconds. xai/grok-imagine-video edits any video up to 8.7 seconds, including an MP4 you send to POST /v1/videos/edits or /v1/videos/extensions.
In workspace chat
Pick a video model and choose the quality, shape and length under the message box. Attach an image to animate it. Each finished video has Edit and Extend buttons, and you can attach an MP4 of your own to edit. You can leave the chat while a video is generating; it keeps going.
Billing
You pay what the provider charges. While a video generates, the most it could cost is held from your balance, and released when it finishes. A video that fails is free. The one exception: xAI charges for a video its moderation blocks, so that one is billed, and the error says so.
Prompts can be anonymized like chat messages, and always are for workspace members with data protection on. Videos are stored encrypted: API videos for 30 days, chat videos as long as their chat.
See Video generation for every option.