LLM Model Providers

DevJock reaches models from major labs through a single gateway. You do not integrate providers one at a time. The gateway exposes a broad catalog behind one interface, so agents and chat sessions can use different models without separate credentials or wiring per provider.

Choosing a model

You can set the model at three levels:

• Platform default. The platform has a default model that agents inherit. Set it once and every agent that does not override it follows along. When you want to move the whole fleet to a newer model, you change one setting instead of editing each agent.
• Per agent. An agent is a versioned prompt plus a model choice. Pin an agent to a specific model when its prompt is tuned for that model's behavior, or leave it on the default to inherit changes automatically.
• Per call. The model can be overridden at call time. This is useful for routing a single request to a larger or smaller model without changing the agent definition.

Model selection is also available in chat, so you can pick the model for a session from the catalog.

Zero-data-retention options

The gateway includes zero-data-retention options. Where a model is configured for zero data retention, prompts and completions are not stored by the provider after the request completes. Availability depends on the provider and model, so check the catalog entry when data handling is a requirement for your workload.

Where to configure it

Set the platform default and per-agent model in the Developer area under Agents. Browse the available catalog from the model picker in chat and in the agent editor. For the exact request fields to override a model per call, see the live interactive AI/Agents API reference reachable from the app's user menu.