Skip to main content
Glama

Generate Video

sirv_generate_video

Generate a video from a text prompt and images, choosing a model, duration, resolution, and optional audio to create new footage ready for download.

Instructions

Make a video from a prompt and images. Pick model by the input you have:

  • gemini-omni-flash (default): a first frame in image_url. 3 to 15 seconds (default 8); 360p, 720p, 1080p or 4k.

  • gemini-omni-flash-ref: 1 to 10 reference images in image_urls. The same lengths and resolutions.

  • ltx: image_url. 6, 8 or 10 seconds; 1080p (default), 1440p or 2160p.

  • kling: image_url. 5 or 10 seconds.

  • h3-max: image_url, or the prompt alone. 5, 10 or 15 seconds; 480P or 768P (default). duration is the length in seconds as text; a length that the model does not offer returns an error. generate_audio adds sound with ltx and kling. The retired veo31 and veo31ref ids run on the gemini models.

Do not use it to make an existing video sharper: use sirv_upscale_video. Returns video_url. The request can take up to 10 minutes: keep the connection open for that long.

Credits: spends workspace credits. The cost depends on the model and options.

Tool call template

Replace each placeholder with a value matching the input schema. Only required parameters are included; add optional parameters as needed.

{
  "id": 1,
  "jsonrpc": "2.0",
  "method": "tools/call",
  "params": {
    "arguments": {
      "prompt": "<prompt>"
    },
    "name": "sirv_generate_video"
  }
}

Input Schema

NameRequiredDescriptionDefault
modelNoVideo model. The retired veo31 and veo31ref run on gemini-omni-flash and gemini-omni-flash-ref. Default: gemini-omni-flashgemini-omni-flash
promptYesThe video to make, in words, up to 2000 characters
durationNoLength in seconds, as text, such as '6'. gemini-omni-flash takes 3 to 15. Default: the model's own
image_urlNoPublic URL of the first frame. Needed for gemini-omni-flash, ltx and kling; optional for h3-max; not used by gemini-omni-flash-ref
image_urlsNo1 to 10 public image URLs that keep the subject the same. Needed for gemini-omni-flash-ref; not used by other models
resolutionNoOutput resolution. Default: the model's own
aspect_ratioNoWidth to height of the video. Default: the model's own
generate_audioNotrue adds a sound track. Default: false

Schema Changelog

Changes observed during successful MCP inspections.

  1. Addedv0.1.1

TDQS

A4.9/5.0
Behavior5/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

Goes well beyond the annotations (readOnlyHint=false, openWorldHint=true, idempotent=false, destructive=false) by disclosing credit spend, a long-running request that can take up to 10 minutes with an instruction to keep the connection open, and that it returns video_url. It also documents failure behavior — 'a length that the model does not offer returns an error' — and that retired veo31/veo31ref ids still run on the gemini models.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness4/5

Is the description appropriately sized, front-loaded, and free of redundancy?

Front-loaded with the core action, then organized as a scannable per-model list, closing with the alternative tool, return value, latency, and cost. Dense but nearly every clause carries decision-relevant information; the only waste is a slight restatement of the veo31 retirement that also appears in the model schema description.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness5/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

For an 8-parameter, non-idempotent, credit-spending generation tool with no output schema, the description covers input mapping, per-model constraints, cost, latency, return value, and error behavior. An agent has everything needed to pick a model and call it correctly without further exploration.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters5/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

Schema coverage is already 100%, yet the description adds real meaning the schema cannot express: per-model allowed durations (ltx 6/8/10, kling 5/10, h3-max 5/10/15, gemini 3–15), per-model resolutions, and which image field each model requires or ignores. It also clarifies that generate_audio only applies to ltx and kling, a conditional constraint absent from the flat boolean in the schema.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose5/5

Does the description clearly state what the tool does and how it differs from similar tools?

Opens with a specific verb+resource ('Make a video from a prompt and images') and then breaks the tool down by its five model modes, so an agent can tell exactly which capability set it is invoking. It is clearly distinguished from sibling generation tools (image, 3d model, depth map) by naming video output and by explicitly excluding upscaling.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines5/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

Provides explicit model-selection guidance keyed to the input the caller already has ('gemini-omni-flash-ref: 1 to 10 reference images in image_urls', 'h3-max: image_url, or the prompt alone'), which is the core routing decision for this tool. It also states an exclusion with the correct alternative: 'Do not use it to make an existing video sharper: use sirv_upscale_video.'

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.