ExLlama
Memory-efficient inference library for quantized Llama models
ExLlama is a standalone Python/C++/CUDA implementation for running quantized GPTQ Llama models with low VRAM use on modern GPUs. It is the predecessor to ExLlamaV2 and focuses on fast, memory-efficient local inference.
Key features
- Low VRAM GPTQ inference
- CUDA-accelerated
- Standalone library
- Fast token generation
Strengths
- Released under the MIT license
- Active community (2.9k GitHub stars)
- Written in Python
ExLlama replaces
Compare ExLlama
10 head-to-head comparisons.
Similar self-hosted ai apps
OpenClaw
Self-Hosted AIThe AI that actually does things
Hermes Agent
Self-Hosted AIThe AI agent that grows with you
OpenCode
Self-Hosted AIThe open source AI coding agent
Hugging Face Transformers
Self-Hosted AIState-of-the-art machine learning model library
Replaces OpenAI API
Langflow
Self-Hosted AIVisual framework for building AI agents and RAG pipelines
Replaces Vertex AI Agent Builder
Dify
Self-Hosted AIOpen-source platform for building production LLM apps
Replaces OpenAI Assistants, Vertex AI Agent Builder