Skip to main content
RelayHub gives your team the AI models of five providers. The models are available through the shared platform keys, through your own keys, or through custom endpoints.

​
OpenAI

GPT-5.6 Luna is made for short tasks. For questions about very large documents, use GPT-5.6 Terra, a Sol model or a Gemini model.

​
Anthropic

​
Google

​
xAI

The xAI models appear in the model picker only when your instance has an xAI platform key, or when your organization added an xAI key on the Providers screen.

​
Paper Instruments

Ultramarine 1 Flash appears in the model picker only when your instance has a Paper Instruments platform key, or when your organization added a Paper Instruments key on the Providers screen.

​
Reasoning and vision

All the models in the tables accept images. All the models except Grok Build 0.1 have a reasoning control. Some models always reason. For these models, the reasoning control gives only depth levels and has no “off” choice: GPT-6 Astra, GPT-6.1 Sol, Claude Fable 5.1, Claude Fable 5, Claude Opus 5.5, Gemini 4 Argon, Gemini 3.8 Flash, Gemini 3.7 Flash, Gemini 3.1 Pro, Grok 4.7, Grok 4.6, Grok 4.5 and Ultramarine 1 Flash.

​
Model selection in chat

Each user selects a model in the model picker of the chat. In Dual Chat, the user selects two models and compares the answers side by side. The models in the picker depend on the setup of your organization:
  • Platform keys: the models of OpenAI, Anthropic and Google, and of xAI and Paper Instruments when your instance has their keys.
  • Your own keys: your key for each provider where you added one. The other providers use the platform keys.
  • Custom endpoints: the models that you imported on your endpoint. If fallback is on, the platform models are also available.

​
Features by provider

Ultramarine 1 Flash of Paper Instruments gives chat, reasoning and image input.
RelayHub makes the embeddings for document indexing and search with an OpenAI or a Google model. If your team chats with Anthropic or xAI models, RelayHub uses a different provider for the embeddings.

​
Image generation models

Image generation on the Create canvas has its own list of models: OpenAI Images 2.6 Flare, OpenAI Images 2.6 Sunburst, OpenAI Images 2.5 Flare, OpenAI Images 2.5 Sunburst, OpenAI Images 2.0, Nano Banana, Nano Banana Pro, Nano Banana 2 Lite, Nano Banana 2.1, Grok Imagine, Grok Imagine Pro and Grok Imagine 2.0.

​
Model updates

RelayHub adds new models to the platform when the providers release them. You do not have to do anything. A new model appears in the model picker after the platform update. If you use a custom endpoint, a new model is available after you deploy it on your endpoint and import it with Discover Models.