MoonshotAI/MoBA

efficient attention for long-context LLMs

open-source memory engine for LLMs and AI agents
GibsonAI/memori
Muon optimizer for training large language models
MoonshotAI/Moonlight
audio foundation model for understanding, generation, and conversation
MoonshotAI/Kimi-Audio
universal memory server for sharing context across all your AI apps
supermemoryai/supermemory-mcp
Structured data extraction for LLMs
shcherbak-ai/contextgem
speeds up LLMs with fast key-value caching
LMCache/LMCache
open source platform for building LLM-powered voice agents and assistants
bolna-ai/bolna
run large language models on any device
mlc-ai/mlc-llm
open large multimodal models for vision and language tasks
LLaVA-VL/LLaVA-NeXT