Model Namespacing
Every OpenGateway model ID follows the owner/model-name pattern — stable naming across providers and clean filtering in logs and billing.
Every model ID follows the pattern owner/model-name.
openai/gpt-4o-minianthropic/claude-sonnet-4google/gemini-2.5-pro
The prefix is the company that built the model, not a reseller.
openai/gpt-4o always refers to the model OpenAI ships. This keeps naming
stable as the ecosystem grows.
Why it matters#
- Stable — the same ID means the same model, regardless of which provider region serves it.
- Unambiguous — no guessing whether
gpt-4ois the OpenAI original or a reseller's copy. - Filterable — logs, dashboards, and billing all group by owner for free.
Coming from OpenAI#
A bare gpt-4o becomes openai/gpt-4o. Strip the namespace and routing
breaks — the gateway needs the owner to pick the provider.
Migrate from OpenAI
The three-line migration, including the model prefix.