refactor: update Google GenAI integration and dependencies

[invincible04]

Description

This PR migrates the Mem0 project from the deprecated google-generativeai library to the new unified google-genai SDK. This migration provides better performance, improved API compatibility, and aligns with Google's latest generative AI SDK recommendations.

Changes Made:

• Changed import statements to use the new google.genai module structure
• Updated the API key environment variable from GOOGLE_API_KEY to GEMINI_API_KEY
• Modified embedding and LLM classes to utilize the new client methods for embedding and content generation
• Adjusted test cases to reflect changes in the API and ensure compatibility with the new SDK structure
• Updated the Makefile to replace google-generativeai with google-genai in the installation command

Motivation:

The google-generativeai library is being phased out in favor of the new unified google-genai SDK, which offers:

• Better performance and stability
• Unified interface for all Google AI services
• Improved error handling and type safety
• Active maintenance and future updates

This migration ensures Mem0 stays current with Google's latest AI SDK recommendations and provides users with a more reliable experience.

Fixes # (no specific issue, but improves dependency management)

Type of change

• [x] Refactor (does not change functionality, e.g. code style improvements, linting)
• [ ] Bug fix (non-breaking change which fixes an issue)
• [ ] New feature (non-breaking change which adds functionality)
• [ ] Breaking change (fix or feature that would cause existing functionality to not work as expected)
• [ ] Documentation update

How Has This Been Tested?

• [x] Unit Test - All existing Gemini-related tests have been updated and pass successfully
  • tests/llms/test_gemini_llm.py - 2/2 tests passing
  • tests/embeddings/test_gemini.py - 1/1 tests passing
• [x] Code Quality Checks - All linting and formatting checks pass
  • Ruff linting: ✅ No issues found in modified files
  • Code formatting: ✅ Applied and verified

Test Configuration:

• Python 3.12.9
• Windows 10 environment
• All dependencies installed via updated Makefile

Checklist:

• [x] My code follows the style guidelines of this project
• [x] I have performed a self-review of my own code
• [x] I have commented my code, particularly in hard-to-understand areas
• [ ] I have made corresponding changes to the documentation (no docs changes needed for internal dependency migration)
• [x] My changes generate no new warnings
• [x] I have added tests that prove my fix is effective or that my feature works (existing tests updated)
• [x] New and existing unit tests pass locally with my changes
• [ ] Any dependent changes have been merged and published in downstream modules (not applicable)
• [x] I have checked my code and corrected any misspellings

Additional Notes:

This is a non-breaking change that maintains full backward compatibility while upgrading to the newer, more robust SDK. Users will benefit from improved reliability and performance without needing to change their existing code or configurations.

The migration has been thoroughly tested and all existing functionality continues to work as expected.

[CLAassistant]

All committers have signed the CLA.

[Dev-Khant]

Hey @invincible04 Can you please resolve the conflicts? Then we can run the tests and merge it, meanwhile I'll review it.

[Dev-Khant]

Hey @invincible04 I actually checked we have already migrated to new library.

• For LLMs: https://github.com/mem0ai/mem0/blob/main/mem0/llms/gemini.py#L6
• For Embedding Model: https://github.com/mem0ai/mem0/blob/main/mem0/embeddings/gemini.py#L4

So closing this PR.

[invincible04]

@Dev-Khant

Can you check the current implementation, we are passing the system prompt in the user prompt, mentioning that it is the system prompt. This is not the good approach. Current Implementation

    def _reformat_messages(self, messages: List[Dict[str, str]]) -> List[types.Content]:
        """
        Reformat messages for Gemini using google.genai.types.

        Args:
            messages: The list of messages provided in the request.

        Returns:
            list: A list of types.Content objects with proper role and parts.
        """
        new_messages = []

        for message in messages:
            if message["role"] == "system":
                content = "THIS IS A SYSTEM PROMPT. YOU MUST OBEY THIS: " + message["content"]
            else:
                content = message["content"]

            new_messages.append(
                types.Content(role="model" if message["role"] == "model" else "user", parts=[types.Part(text=content)])
            )

        return new_messages

In gemini API, we have the config to set the system prompt, which I have implemented My Implementation in my refactored codebase.

Please look into it and use the best approach in the project, so that in future, there can be possibility for many more features integration, like context caching etc.

[Dev-Khant]

Hey @invincible04 Yes you are correct this should be fixed. Can you please raise a new PR with this change for gemini?

[invincible04]

Hi @Dev-Khant , I would kindly request you to please look the code before merging it to the main branch. In this file: https://github.com/mem0ai/mem0/blob/main/mem0/embeddings/gemini.py If you see the API key is passed as "api_key" string, instead of api_key variable. This way, it will never be initialised no matter what if the API is correct or not.

api_key = self.config.api_key or os.getenv("GOOGLE_API_KEY")
if api_key:
      self.client = genai.Client(api_key="api_key")
else:
      self.client = genai.Client()

Also, in the https://github.com/mem0ai/mem0/blob/main/mem0/llms/gemini.py, the error code says to download google-generativeai deprecated library instead of google-genai, even though we are using google-genai library.

[Dev-Khant]

Hey, Missed this will take proper look next time currently have too many things going on. Fixing it now.

[invincible04]

You have already fixed the implementation, so i won't be creating a new PR.

Will check for other bugs in the repo!