Skip to main content
RelayHub gives your team access to the latest AI models from four leading providers. Models are available through the platform’s shared keys, your own BYOK keys, or custom endpoints.

​
OpenAI

RelayHub uses the OpenAI Responses API for all GPT-5 series models, ensuring access to the latest capabilities including native tool use and structured outputs.
GPT-5.6 Luna is tuned for speed and cost on short tasks. For questions over very large documents, pick GPT-5.6 Terra, GPT-5.6 Sol, or a Gemini model instead.

​
Anthropic

​
Google

Google Gemini models support the longest context windows, making them well-suited for analyzing large documents or lengthy conversation histories.

​
xAI

xAI models appear in the model picker only when your instance has an xAI platform key or your organization has added an xAI key in the Provider Hub.

​
Model Capabilities

Reasoning (extended thinking) and vision (image input) support varies by model: Some models always reason and cannot have it switched off: Claude Fable 5, Gemini 3.1 Pro, Grok 4.5, and Grok Build 0.1. For those the reasoning control offers depth levels only.

​
Model Selection in Chat

Users select their preferred model from the model picker at the top of any chat session. In Dual Chat mode, users select two different models (from the same or different providers) to compare responses side by side. The available models in the picker depend on your organization’s configuration:
  • Platform keys: All models from all four providers are available
  • BYOK keys: Your keys are used for the providers where you added one; other providers continue on platform keys
  • Custom endpoints: Only the models discovered or configured on your endpoint (plus platform models if fallback is allowed)

​
Feature Support by Provider

Embedding generation (used for document indexing and semantic search) is handled by OpenAI or Google embedding models. If you use Anthropic or xAI as your primary chat provider, RelayHub will use a separate embedding provider for indexing tasks.

​
Image Generation Models

Image generation on the Create canvas uses a separate catalog: OpenAI Images 2.0, Google Nano Banana 2 and Nano Banana Pro, and xAI Grok Imagine and Grok Imagine Quality.

​
Model Updates

RelayHub’s model catalog is updated regularly as providers release new models. Updates happen on the platform side — no action is needed on your part. New models automatically appear in the model picker after a platform update. If you use a custom endpoint, new models become available as soon as you deploy them and re-run model discovery.