One Workspace, Five Providers: How Model Switching Works

RG AI isn't betting on a single model provider long-term, and that's deliberate. Every provider is good at something different, whether that's cost, speed, reasoning depth, or context length, and the right choice often depends on the task in front of you, not on which company you happened to sign up with first. Chat runs free on one built-in model, and connecting your own providers unlocks the rest.

Picking a model

The model selector in the composer lists every provider you've connected. Pick one, send a message, and RG AI routes that request directly to the matching API: Anthropic's Messages API for Claude, OpenAI-compatible endpoints for ChatGPT, Kimi, and DeepSeek, and Google's Generative Language API for Gemini.

• Claude: strong reasoning and long-context work.
• ChatGPT: broad general-purpose performance.
• Gemini: fast, multimodal-friendly responses, genuinely free tier.
• Kimi: long-context conversations at a lower cost.
• DeepSeek: strong coding output, free to start, extremely cheap at scale.

Why Gemini or DeepSeek are the easiest way to start

Gemini or DeepSeek are the obvious first stop: both hand out a free key with no card required, and respond fast enough that you won't notice you're not on a paid model. From there, connecting a second or third provider takes under a minute from the same Settings panel. Chat is also free on RG AI's own built-in model any time you'd rather not use a key at all.