Understand model creators, identifiers and tags

A model creator publishes or authors the model. A provider operates the infrastructure that runs your request. One model can be available from several providers, with different prices and observed performance.

Browse by model, compare by provider#

1. Open Models and choose a model family.
2. Read the creator name and follow its profile to see other attributed models.
3. Select the exact tag or variant you need.
4. Compare the execution providers and their applicable input, output and cache prices.
5. Copy the model identifier and request example for your endpoint.

Keep the creator in the model identifier#

{
  "model": "openai/gpt-5.4",
  "provider": {
    "only": [
      "shareai"
    ]
  },
  "messages": [
    {
      "role": "user",
      "content": "Hello"
    }
  ]
}

This example asks the ShareAI execution provider to run an OpenAI model. Availability and key permissions still apply. Do not add an execution-provider prefix such as shareai/ to choose the provider; use provider in the request body.

Tags remain distinct#

A family page groups related tags. A different size, revision or quantization remains a separate model choice. Provider fallback keeps the requested model; it does not silently choose another tag or model family. A fine-tune may name its own author rather than the author of its base model.

Attribution and claims#

An unclaimed creator profile records public attribution. It does not mean a company account has been verified, accepted into a program or granted model royalties. See claiming a creator profile.

Next#

Choose execution providers · Provider policy reference