> ## Documentation Index > Fetch the complete documentation index at: https://docs.tryblend.ai/llms.txt > Use this file to discover all available pages before exploring further. # Video Gen > Generates a video from a prompt and optional reference media using the model you select. The Video Gen node runs a video model on a prompt and produces a video for the next node to use. You pick the model and write the prompt. Depending on the model you choose, the node can also accept reference media, such as a starting image, to guide the generation. Both the prompt and the model's parameters can be set on the node or fed in from upstream nodes. ## Inputs Input handles are on the **left** of the node. The exact set depends on the model you select — the node shows a handle for each of that model's available parameters. | Handle | Type | Required | Notes | | ---------------------------- | ------------------- | -------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ | | Prompt | `text` | No | The text description of the video to generate. Type `{{variable}}` tokens to create one input handle per variable. | | Reference / image parameters | `image` (or `file`) | No | Appears when the selected model accepts reference media, such as a start image. Image-only parameters expose an `image` handle; parameters that accept several media kinds use the generic `file` handle. These accept **multiple** connections. | | Other model parameters | varies by model | No | Additional parameters the selected model exposes (e.g. a numeric or select setting) each get their own handle so you can drive them from upstream. | | Model | `model` | No | Optional handle to set the model from an upstream node instead of picking it on the node. | ## Outputs Output handles are on the **right** of the node. | Handle | Type | Notes | | ------ | ------- | ----------------------------------------------------- | | Video | `video` | The generated video. Feeds the next node in the flow. | One video generation call produces one output item. If the Video Gen node receives multiple input items from a Batch or structured list, it runs once per item and produces one output item per run. ## Settings * **Model** — the video model that runs. Filtered to models that produce video output. * **Prompt** — the text instruction the model generates from. A `{{variable}}` token exposes a dedicated text handle so upstream values can fill that part of the prompt. * **Connected parameters** — any extra inputs the selected model exposes (reference media, aspect ratio, duration, and similar) appear as fields on the node. The exact set changes with the model. * **Advanced Settings** — less-common model parameters under a collapsible section. The available parameters depend on the selected model. Switching models keeps any compatible parameter values and fills in defaults for the new model's parameters. ## When to use Reach for Video Gen whenever a step should produce a video — turning a text prompt into a clip, or animating a reference image the upstream flow created. Pick the model that matches the style, length, and inputs you need. ## Connects to * **Upstream** — [Input](/workflow/nodes/input) nodes feeding the prompt; an [LLM](/workflow/nodes/llm) node producing a refined prompt; an [Image Gen](/workflow/nodes/image-gen) node producing a reference/start image. * **Downstream** — a [Result](/workflow/nodes/result) node to surface the video as a workflow output, or any node that accepts a `video` (or `file`) input. ## Notes / limits * The output is a single `video` handle labeled **Video**. * Video Gen can run by itself from the node's Run action when its required inputs are available. * Video generation typically takes longer and costs more than text or image steps; runs can take a while to complete. * A Batch, multi-select Choose Output, or LLM structured list makes Video Gen run once per incoming item. See [Lists & single values](/concepts/lists-and-values). * Connection compatibility follows the standard rules: exact-type matches connect, `any` connects to anything except `tool-list`, and media types can feed a `file` target. * As with the rest of the canvas, no cycles or self-connections. ## Next steps How nodes, handles, and connections fit together on the canvas. Generate an image — often the reference a Video Gen node animates.