LLM Providers

TeleMem works with any OpenAI-compatible endpoint: set provider: openai and point openai_base_url at your service. Ready-to-use config examples ship in config/:

Provider	Config file	LLM	Embeddings	Notes
Ollama (fully local)	config.ollama.yaml	any local model (e.g. qwen3:8b)	nomic-embed-text, local	No API key, no cloud
DeepSeek	config.deepseek.yaml	deepseek-chat / deepseek-reasoner	external (e.g. OpenAI)	export DEEPSEEK_API_KEY=...
Moonshot (Kimi)	config.moonshot.yaml	kimi-k2-0905-preview	external (e.g. OpenAI)	.cn and .ai endpoints
MiniMax	config.minimax.yaml	MiniMax-M3 (512K context)	external (e.g. OpenAI)	temperature must be in (0.0, 1.0]
Local Qwen (vLLM etc.)	config.yaml	qwen3-8b	qwen3-8b-embedding	the configuration used in the Tech Report
OpenAI	(default)	Memory() with no config	OpenAI	needs OPENAI_API_KEY

Usage

from telemem.utils import load_config
import telemem as mem0

config = load_config("config/config.ollama.yaml")
memory = mem0.Memory(config=config)

Or via the environment for the bundled examples:

TELEMEM_CONFIG=config/config.deepseek.yaml python examples/quickstart.py

Fully local (Ollama)

ollama pull qwen3:8b
ollama pull nomic-embed-text
TELEMEM_CONFIG=config/config.ollama.yaml python examples/quickstart.py

Everything — LLM, embeddings, FAISS vector store, JSON metadata — runs on your machine.

Adding a provider

Contributions of new provider configs are very welcome — copy an existing config/config.<provider>.yaml, add offline tests like tests/test_providers.py, and open a PR. See issue #9.

Dimension consistency

If your embedder's output dimension differs from the default, pin it in both places:

embedder:
  config:
    embedding_dims: 768          # what the model outputs
vector_store:
  config:
    embedding_model_dims: 768    # what the FAISS index expects