Models

Runbase provides access to top image and video generation models from multiple providers through a single API.

Browsing models

Visit the Models page to see all available models with pricing, capabilities, and example outputs.

Each model has a dedicated detail page with:

• Playground — test the model in your browser before writing any code
• Pricing — per-run or per-second cost
• Examples — sample outputs with prompts
• API reference — model-specific input parameters and code samples

Playground

Every model page includes an interactive Playground at the top. Sign in, type a prompt, and click Run model to see the result in seconds. This is the quickest way to evaluate a model's quality and speed before integrating it into your application.

Model ID format

Every model has an ID in the format provider/model-name, for example:

• openai/gpt-image-2 — image generation
• bytedance/seedance-2 — video generation
• google/veo-3 — video generation

The full list of available models and their IDs is on the Models page. Models are added regularly — check there for the latest.

Input parameters

Each model defines its own input schema. Common parameters include:

• prompt (string) — text description of what to generate
• aspect_ratio (string) — output dimensions, e.g. "1:1", "16:9"
• quality (string) — generation quality tier
• image_urls (string[]) — URLs of reference images (where supported)

See each model's detail page for the full input schema and available options.

Using a model

Pass the model ID and input object to the create run endpoint:

curl -s https://runbase.net/api/v1/runs \
  -H "Authorization: Bearer $RUNBASE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "bytedance/seedream-5-lite",
    "input": {
      "prompt": "A mountain landscape at sunset, oil painting"
    }
  }'

If you pass an invalid model ID, you'll get a 400 model_not_found error. If the input doesn't match the model's schema, you'll get 400 invalid_input with a description of what's wrong.

Image vs. video models

• Image models return one or more image URLs in output.urls. Generation time varies by model, provider load, and input complexity.
• Video models return a video URL in output.urls. Generation time varies by model, duration, resolution, and provider load.

The API shape is identical for both — the only difference is what the output URLs point to.