mem0-embeddings-litellm-patch

This patch adds support for embedding model providers via LiteLLM to the mem0 framework.

✨ What It Does

• Integrates nearly all providers supported by LiteLLM as embedding backends in mem0
• Enables use of high-performance providers like VoyageAI, Mistral, Groq, and more
• Drop-in replacement for the existing embedding logic

🔧 Installation

You can install the patch via pip:

pip install mem0-embeddings-litellm-patch

This will patch the necessary mem0.embeddings modules automatically.

Note: Make sure mem0 and litellm are installed as dependencies. This package does not install them implicitly.

🧠 Requirements

• Python >= 3.8
• mem0 >= 0.1.0
• litellm >= 1.0.0

💡 Usage

After installing this patch you can use all embedding providers available via litellm inncluding those currently not supported via mem0 natively.

📢 Why This Exists

The mem0 maintainers have not yet merged support for LiteLLM-based embeddings, despite it being a fast, extensible abstraction layer. This patch bridges the gap until (or if) native support is added upstream. No need to fork and maintain a full project if you can just maintain the patch files instead am i right? :D No need to fork and maintain a full project if you can just maintain the patch files instead am i right? :D

📬 Feedback / Contributing

Feel free to fork or open issues. If the mem0 team integrates this feature officially, this package may be deprecated in favor of upstream support.

Licensed under the MIT License.