What to check before adding image or video generation
Compare modality, endpoint support, billing and output handling before adding an image or video model to your application.
Updated October 2, 2026 · G0I documentation team
Do not confuse image understanding with generation
A model that accepts image input is not necessarily an image generator. Confirm the exact task supported by the model and endpoint. The catalog’s type and description can help shortlist candidates, but your integration should validate the requested input and output format.
Inspect the billing unit
Token rates do not describe every media workflow. Image size, duration, output format or a job-specific billing rule may matter. Review the documentation and current account information before estimating the cost of a batch of generations.
Plan for asynchronous work
When an endpoint returns a job rather than the completed asset, record its identifier and handle pending, completed and failed states. Use bounded polling and avoid creating duplicate jobs after a timeout unless you know the first request did not succeed.
Validate and store output safely
Check returned asset types and sizes, enforce appropriate retention, and avoid exposing private prompts or files in public logs. Follow the applicable policies and review generated content before publishing it. Availability of a model can change; design the UI to explain failures clearly.
Continue with the documentation
Compare published models, review access plans, or start with the API quickstart.