> ## Documentation Index
> Fetch the complete documentation index at: https://docs.paradime.io/llms.txt
> Use this file to discover all available pages before exploring further.

# Gemini, GPT-5.6, Grok, Kimi, and Qwen come to DinoAI

> DinoAI's model picker now spans Gemini, GPT-5.6, Grok, Kimi K2 Thinking, and Qwen3 alongside Claude, grouped by provider, each at its own credit rate.

*August 26, 2026*

<Frame>
  <img src="https://mintcdn.com/paradime-docs/Z2HXRo1EOxZRJjC-/images/dinoai-new-models-providers.jpg?fit=max&auto=format&n=Z2HXRo1EOxZRJjC-&q=85&s=d83c5f9375e1d4e6911f5870d4dc8103" alt="New models in DinoAI: Paradime connected to the Gemini, OpenAI, xAI, Kimi, and Qwen provider logos" width="1600" height="900" data-path="images/dinoai-new-models-providers.jpg" />
</Frame>

DinoAI is no longer Claude-only. The model picker now covers six providers: Google's Gemini 2.5 and 3.x, OpenAI's GPT-5.6 family (Sol, Terra, and Luna), xAI's Grok 4.x, Moonshot AI's Kimi K2 Thinking, and Qwen3 235B, alongside the existing Claude models.

Models are grouped under their provider in the picker, and each one bills at its own credit rate, so you can match the model to the job: a flash-class model for quick questions and everyday edits, a frontier model for the gnarly refactor. Every request routes through Paradime's LLM gateway with the same session experience, tools, and [credit metering](/products/dino-ai/dinoai-credits) regardless of provider.

See the full lineup and when to use each model in [Supported models](/products/dino-ai/supported-models).

Shipped in Paradime v6.9.0.


## Related topics

- [Supported models](/products/dino-ai/supported-models.md)
- [What is DinoAI?](/guides/foundations/what-is-dinoai.md)
- [Paradime MCP server: connect AI clients to your data](/products/dino-ai/mcp-server/index.md)
- [Fix failed Bolt runs with DinoAI on the run page](/changelog/2026-08-26/bolt-run-dinoai-tab.md)
- [Budget & Consumption for DinoAI credits](/changelog/2026-07-29/budget-and-consumption.md)