Generate Grok Imagine clips from an agent with the generate_video tool
Hosted MCP exposes generate_video, which replaced video-router_create. How an agent should call it for grok-imagine-video-1.5 and what to read first.

Yes, an agent can make Grok Imagine clips on Sume through hosted MCP: the paid tool is generate_video, which replaced the retired alias video-router_create. The Grok row, grok-imagine-video-1.5, is image-to-video only, so the call must carry a first frame, and the tool's own contract is the thing to read before you guess any field. Ask for it with tools_schema and the tool name.
The order an agent should follow
The MCP docs split tools into discovery, free reads and paid writes. tools_list, tools_schema and mcp_health are for discovery. generate_video is in the paid group with generate_image and music_create, and a paid write carries an idempotency_key, so a retried call does not make a second job.
| Step | Tool | Why |
|---|---|---|
| 1 | tools_schema with generate_video | Read the exact input contract |
| 2 | generate_video | Submit with an idempotency_key and the Grok model id |
| 3 | jobs_wait | Wait on the returned job id |
| 4 | A fresh call only with a new key | Never reuse a key with a changed body: that is a 409 |
An instruction to paste
Give the agent the constraints in its instruction, since the schema cannot say which model is right for the job:
Make one 6-second clip with model grok-imagine-video-1.5.
It is image-to-video only: use the supplied still as the first frame.
Resolution is 480p or 720p, duration 4 to 15 seconds.
First call tools_schema for generate_video and follow it exactly.
Send an idempotency_key. If the call fails, retry with the same key.
Wait on the job id; do not submit a second job.Cost and limits
Cost is the same as over HTTP: provider list times 1.25, ceiled to the cent per job. At $0.0125 a second, 6 seconds is $0.075, billed as $0.08. xAI's models page also lists a Lite video model, which is not in Sume's documented catalog.
Sources
Related posts
More in Developers
- Can the Sume Image API output SVG? Not today; formats you get
No: output_format svg is in the schema for parity, but no catalog model advertises it, so the request is rejected. Sume returns png, jpeg or webp.
- GET /v1/balance returns available_credits: a TypeScript gate
Sume's balance read returns whole credits in available_credits, a funded or empty state, and expiring_soon_credits. Gate a job on it in TypeScript.
- Go 1.27.2 synctest: test a 30-second Sume poll loop in milliseconds
Test a Go poll loop for Sume video jobs with testing/synctest: 90 seconds of fake sleeps and a 20-minute deadline pass in under a second, no sockets.
- Image edit reference limits on Sume: 16 on GPT Image 2.5, 10 elsewhere
ChatGPT Image 2.5 takes up to 16 input_references; most edit rows take 10, Ideogram 4.5 takes 5. Imagen, Recraft and Qwen Image Max take none.
Written by Sume