Gemini model list in v0.25

[systemz64]

Hi
I updated to v0.25 recently. I already had a chat setup with gemini 2.5 pro - all still works. But due to the end of support I wanted to update this to use gemini 3-flash. However, in the model setup, I still see the old list of models - I dont see any of the newer gemini models. If I try and select any - even if I try and re-apply 2.5 pro - I get an error saying its not available? Just to understand - would the latest ragflow version not have been updated for the latest gemini models?

[systemz64]

Hi All

still no luck - I have removed gemini from the model providers and then readded - I still cannot chnage the model on an existing chat setup - see attached screenshot - any advice would be appreciated

[Ganesh-403]

@systemz64 From what I can tell, this is likely a combination of cached model metadata and provider support in RAGFlow v0.25, rather than an issue with your Gemini API key.

A couple of things to check:

1. Can you create a new chat assistant using the newer Gemini model?

   • If a newly created assistant shows the updated models but an existing one does not, it may indicate that the model selection for existing assistants isn't automatically refreshed after upgrading or changing providers.

2. Verify the exact model ID you're trying to use.

   • Google has deprecated and renamed several Gemini models over time. Using an older model ID will result in an error even if your API key is valid. You can compare against Google's current supported model names here: :contentReference[oaicite:0]{index=0}

3. Check whether RAGFlow v0.25 bundles a static model list.

   • I couldn't find documentation confirming that v0.25 dynamically fetches the latest Gemini models from Google's API. If the list is bundled with the release, newer Gemini models wouldn't appear until RAGFlow updates its provider integration. :contentReference[oaicite:1]{index=1}

A few questions that would help narrow it down:

• Are you using the Google AI (Gemini Developer API) or Vertex AI provider?
• Which exact model are you trying to select (e.g. gemini-2.5-flash, gemini-2.5-pro, etc.)?
• Are you running the official Docker image for v0.25, or did you upgrade from an earlier version?

If the issue only affects existing chat setups while newly created ones work correctly, I'd lean toward this being a RAGFlow UI/configuration issue rather than a Gemini API problem.

If this solves your problem, feel free to mark it as the accepted answer so others can find it easily.