Video API
Video generation API: coming soon
Prompt-to-video and image-to-video are not available in Adlark yet, in the studio or the API. This page explains what is planned and what you can build today with the endpoints that are live.
Current status
There is no live video generation endpoint. Calls to generate video will not work until we launch it. We would rather say so plainly than sell credits for something that is not ready. Everything else in the API (images, copy, UGC scripts, analysis, captions, voiceover, jobs and credits) is live.
What is planned
The plan is video clips from a text prompt or a starting image, using Google's Veo 3.1 models (Fast and Quality) through the Gemini API. We are not affiliated with Google. Planned clip lengths are 4, 6 and 8 seconds. In the catalog today, the planned prices are 32 credits per second for Veo 3.1 Fast and 84 credits per second for Veo 3.1 Quality in the studio, with API calls at plus 10% as usual. These may change before launch; the pricing page will always show the current numbers. Details on the model are on the Veo page.
It will be asynchronous
Unlike the other endpoints, video takes longer than a normal HTTP request should wait. Expect a pattern like this: submit a job, get a job id, then poll the jobs endpoint until it is done. The jobs endpoint exists today, so you can build your client around it now.
# Works today for any job: check status and output
curl https://adlark.alexshipit.com/api/v1/jobs/JOB_ID \
-H "Authorization: Bearer adl_YOUR_KEY"Build the rest of the pipeline now
- Write the script with
POST /api/v1/ugc-script(15, 30 or 60 seconds). - Make a storyboard image per scene with
POST /api/v1/imagesat 9:16 or 16:9. These frames are also natural starting images for image-to-video later. - Generate narration with
POST /api/v1/voiceover. - Edit with your own footage, then caption it with
POST /api/v1/captions.
When video launches, it slots in at step 2 without changing the rest. Code examples: Python and JavaScript.
Errors to handle now
Build in handling for 401, 402, 429 and 503 with { error } bodies. A 503 is also what a switched-off model returns, so a client that handles it gracefully will behave well during any rollout.
For honest expectations about AI video ads in general, read AI video ads: what to expect.
Questions
+Can I generate video with the API today?
No. Video generation is coming soon. The image, copy, script, analysis, captions and voiceover endpoints are live.
+Will video be available over MCP too?
That is the plan once video launches, using the same key and credits.
+Will prices be the same as listed now?
The catalog lists planned prices, but they can change before launch. The pricing page shows current numbers.
Related
Make your next ad today
Pro is $39.99 a month with 3,000 credits. Cancel anytime.