Developer

Build with every model through one clean API.

ModelVia gives developers one place to explore models, test prompts, call text, image, and video routes, and keep request handling consistent as providers change.

Explore modelsGet API access
quickstart.sh
curl https://api.modelvia.ai/v1/chat/completions \
  -H "Authorization: Bearer $MODELVIA_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "openai/gpt/text",
    "messages": [
      { "role": "user", "content": "Summarize this product brief." }
    ]
  }'
9+
available model families
3
model categories
1
API key and billing surface

Model capabilities

One catalog for text, image, and video routes.

Text models

Call language models for chat, reasoning, coding, summarization, extraction, and routing logic.

View routes

Image models

Generate or edit images with text-to-image and image-to-image routes using the same API key.

View routes

Video models

Create short clips from text, image, reference media, video, or audio inputs through asynchronous prediction routes.

View routes

Quick start

Four steps from model choice to result.

These steps are specific to ModelVia: choose a normalized route, use one API key, send the matching request, then read either a direct text response or an asynchronous prediction result.

01

Choose a model route

Browse the model catalog and select a route such as text, text-to-image, or text-to-video.

02

Create an API key

Keep the key on your server and send it in the Authorization header for every request.

03

Send a request

Use chat completions for text, or generateImage and generateVideo for media tasks.

04

Read the result

Text routes return directly. Image and video routes return a prediction id that you poll until complete.

Build with code

Request
API
curl /v1/chat/completions \
  -H "Authorization: Bearer $MODELVIA_API_KEY" \
  -d '{
    "model": "openai/gpt/text",
    "messages": [{ "role": "user", "content": "Summarize this." }]
  }'
Summarize ModelVia in one sentence.
ModelVia gives teams one API surface for routing, testing, and scaling AI model calls.

API surface

The smallest useful reference.

This page intentionally stays focused on the core routes needed to start calling models. Full per-model parameters live on each model route page.

POST
/v1/chat/completions

OpenAI-compatible text requests with model, messages, temperature, and token controls.

POST
/v1/generateImage

Asynchronous image generation and editing using model plus input fields.

POST
/v1/generateVideo

Asynchronous video generation for text, image, video, and audio-driven workflows.

GET
/v1/generateImage/{generation_id}

Polling endpoint for queued media jobs, status, output URLs, and errors.

GET
/v1/generateVideo/{generation_id}

Polling endpoint for queued media jobs, status, output URLs, and errors.

POST
/v1/model/uploadMedia

Upload source images, videos, or audio before submitting reference-based jobs.

FAQ

Basic developer questions.

Do I need different SDKs for each provider?

No. ModelVia keeps provider selection behind normalized routes, so your application can keep one authentication and request flow.

Which model types are supported?

The current catalog focuses on text, image, and video models, including text-to-image, image-to-image, text-to-video, image-to-video, reference-to-video, video-to-video, and audio-to-video routes.

How are media results returned?

Media requests return a prediction id first. Poll the prediction endpoint until the job completes, then read the generated output URLs.