[Feature] [Gemini] Implement Full Google Gemini API Support (UI + Backend)

[HexdoAI]

Parent Epic: #56
Depends on: #61 (This issue introduces the "Provider Type" UI)

🎯 Goal

Implement end-to-end support for the Google Gemini API, allowing users to add it as a custom model provider.

💻 UI Requirements (Frontend)

1. Add "Gemini" as an option to the "Provider Type" dropdown menu (introduced in [Feature] [Azure] Implement Full Azure OpenAI Support (UI + Backend) #61).
2. When a user selects "Gemini", the form must show the following fields:
   • API Key
   • API Address (Optional: Allow users to specify a custom/proxy endpoint instead of the default generativelanguage.googleapis.com)

⚙️ Backend Requirements

1. Modify the "save model" API to store the provider_type: 'Gemini' and its associated fields.
2. The backend logic must check the provider_type for the model.
3. If the provider_type is "Gemini", the system must route the API call to the llm.gemini.js client (this client may need to be created or updated).
4. The llm.gemini.js client must correctly use the API Key and target the API Address (or the default Gemini endpoint if not provided) to make successful API requests.

✅ Acceptance Criteria

• A user can successfully add and save a Gemini API configuration via the UI.
• The system can use this configuration to successfully make calls to the Gemini API.

[JBoschsoft]

Hello, is this issue getting fixed, as it seems impossible to use Gemini on Lemon AI using Docker locally

Error "LLM API call failed, HTTP status: 400, error: [{ "error": { "code": 400, "message": "Invalid JSON payload received. Unknown name "enable_thinking": Cannot find field.", "status": "INVALID_ARGUMENT", "details": [ { "@type": "type.googleapis.com/google.rpc.BadRequest", "fieldViolations": [ { "description": "Invalid JSON payload received. Unknown name "enable_thinking": Cannot find field." } ] } ] } } ]"

In the selection I have labeled with two seperate attempts "Vision" & "Tool" and in the second one "Vision" & "Tool" & "Reasoning"

[thibaultbnf]

Hello, is this issue getting fixed, as it seems impossible to use Gemini on Lemon AI using Docker locally

Error "LLM API call failed, HTTP status: 400, error: [{ "error": { "code": 400, "message": "Invalid JSON payload received. Unknown name "enable_thinking": Cannot find field.", "status": "INVALID_ARGUMENT", "details": [ { "@type": "type.googleapis.com/google.rpc.BadRequest", "fieldViolations": [ { "description": "Invalid JSON payload received. Unknown name "enable_thinking": Cannot find field." } ] } ] } } ]"

In the selection I have labeled with two seperate attempts "Vision" & "Tool" and in the second one "Vision" & "Tool" & "Reasoning"

did you fix this issue ?