Anybody tried to use AzureChatOpenAI to connect Gemini?

[postbird]

Checked other resources

• I added a very descriptive title to this question.
• I searched the LangChain documentation with the integrated search.
• I used the GitHub search to find a similar question and didn't find it.

Commit to Help

• I commit to help with one of those options 👆

Example Code

see description

Description

I'm using the ggemini-2.5-pro-preview-06-05 model, due to my llm provider, I need to connect the llm provider by AzureChatOpenAI in langchain which like below:

But when I build an agent via createReactAgent() method, and invoke it, I got the error:

  **Error: 400 400: Invalid JSON payload received. Unknown name "$schema" at 'tools[0].function_declarations[0].parameters': Cannot find field.**
      at Function.generate (.pnpm/openai@4.98.0_zod@3.25.62/node_modules/openai/src/error.ts:72:14)
      at AzureOpenAI.makeStatusError (.pnpm/openai@4.98.0_zod@3.25.62/node_modules/openai/src/core.ts:462:21)                                                    
      at AzureOpenAI.makeRequest (pnpm/openai@4.98.0_zod@3.25.62/node_modules/openai/src/core.ts:526:24)                                                        
      at processTicksAndRejections (node:internal/process/task_queues:95:5)                                                         
      at async .pnpm/@langchain+openai@0.5.12_@langchain+core@0.3.57/node_modules/@langchain/openai/dist/chat_models.cjs:2337:29                                 
      at async RetryOperation._fn (.pnpm/p-retry@4.6.2/node_modules/p-retry/index.js:50:12)

I am not sure if this way can be work for my scene, so finding some help.(Tried to search a lot of issues and information, no valid items.)

System Info

"@langchain/core": "^0.3.57",
"@langchain/deepseek": "^0.0.1",
"@langchain/langgraph": "^0.3.0",
"@langchain/mistralai": "^0.2.0",
"@langchain/openai": "^0.5.12",

[dosubot[bot]]

AzureChatOpenAI is specifically built for Azure's OpenAI-compatible endpoints and expects payloads and tool/function schemas that match the OpenAI API. Gemini models (like ggemini-2.5-pro-preview-06-05) use a different API and schema, so AzureChatOpenAI can't connect to Gemini models directly—this is why you're seeing the JSON payload error about unknown fields like $schema in function declarations. The error is due to a mismatch between the expected OpenAI/Azure schema and Gemini's schema requirements, not a bug in your code or configuration reference.

If you need Gemini support in LangChain, you should use the dedicated integration for Gemini models, such as ChatGoogleGenerativeAI from the langchain-google-genai package reference. If your provider only exposes Gemini via an OpenAI-compatible API, it must fully match the OpenAI schema—including how tools/functions are declared—which Gemini does not.

If you have to use a custom endpoint or provider, you may need to subclass BaseChatModel and implement the Gemini-specific API contract, similar to how other providers are integrated reference.

Let me know if you want pointers on building a custom integration or have more details about your provider's API!

To reply, just mention @dosu.

---

^{How did I do? Good | Irrelevant | Incorrect | Verbose | Hallucination | Report 🐛 | Other}