PasLLM is finally publicly released!
A high-performance LLM inference engine written in pure Object Pascal.
✨ No Python dependencies
🎯 Advanced 4-bit quantization
🖥️ Cross-platform (Delphi & FreePascal)
⚡ Native performance
Supports Llama 3.x, Qwen, Phi-3, Mixtral, DeepSeek R1 & more!
Open source (AGPL 3.0): https://github.com/BeRo1985/pasllm (synced from my private repo on my on server)
