vLLM is a high-throughput, memory-efficient inference and serving engine for large language models (LLMs). It implements PagedAttention for near-zero waste KV cache management, continuous batching, chunked prefill, prefix caching, CUDA/HIP graphs, speculative decoding, and torch.compile. The system serves OpenAI-compatible APIs (chat completions, completions, embeddings, rerank, speech-to-text), Anthropic Messages API, Cohere API, and custom pooling/classification endpoints. It supports 200+ model architectures including dense, MoE, hybrid SSM/attention, and multimodal models (vision, audio, video). The architecture is a hybrid Python/Rust system: a Python core handles model execution, scheduling, and kernel dispatch, while a Rust frontend provides a high-performance HTTP/gRPC server with middleware for auth, metrics, and load management. Key entry points include the `vllm` CLI (serve, bench, launch), the `LLM` class for offline inference, and the OpenAI-compatible API server. External dependencies include PyTorch, HuggingFace Transformers, FlashAttention, CUTLASS, Triton, NCCL/RCCL, and various quantization libraries. The project is production-grade with extensive CI/CD, Docker images, Helm charts, and deployment integrations for Kubernetes, Ray, SageMaker, and more.
Ollama is a local LLM serving platform written in Go. It provides a REST API (OpenAI-compatible), CLI, TUI chat, and desktop application for running large language models locally. The system supports multiple inference backends: llama.cpp (CPU/CUDA/ROCm/Vulkan) and MLX (Apple Silicon). It includes a model conversion pipeline from HuggingFace safetensors to GGUF format, GPU hardware discovery, and an agent framework with tool execution (bash, file operations, web search). The architecture is a monolith with subprocess management for inference backends. Key entry points include the CLI (main.go), HTTP server (server/routes.go), runner binary (cmd/runner/main.go), and desktop UI (app/ui/app/). The codebase is actively maintained with extensive test coverage and support for dozens of model architectures. While primarily an application, it has significant data/research components in model conversion, tokenization, and speculative decoding research.
GenieX is a production-grade, on-device generative AI inference SDK and CLI developed by Qualcomm for Snapdragon platforms. It provides a C core with a plugin architecture supporting two inference backends: llama.cpp (GGUF models on CPU/GPU/NPU) and QAIRT (pre-compiled bundles on Hexagon NPU). The system includes a Rust-based model manager that downloads, caches, and resolves models from Hugging Face, Qualcomm AI Hub, Docker Hub, and local filesystem, with resumable chunked downloads and progress reporting. It exposes multiple client interfaces: a Go CLI with subcommands for inference, an OpenAI-compatible HTTP server (streaming chat/completions, tool calls), Python bindings with high-level AutoModel factories, Android bindings via JNI/Kotlin, and Go bindings. The architecture is cleanly layered: C SDK → plugin backends → hardware, with a separate model manager process. The codebase demonstrates strong engineering practices: extensive test suites (unit, integration, device-level), CI/CD pipelines, versioned releases, and cross-platform support (Linux ARM64, Windows ARM64, Android). While primarily an application, it includes benchmark infrastructure and quality evaluation tools that could be used for research purposes.
Atomic Chat is a full-featured desktop application for running local large language models and connecting to cloud AI providers. Built with Tauri v2 (Rust backend) and a React/TypeScript frontend, it provides a unified chat interface, agent mode with tool execution, MCP server integration, model hub, and an OpenAI-compatible local API server. The application supports multiple inference backends (llama.cpp, MLX, Apple Foundation Models) and handles model downloads, hardware detection, and agent orchestration. It includes extensive testing, telemetry (PostHog/Sentry), and a plugin architecture for extensibility. The codebase demonstrates clean separation of concerns through a service hub pattern, consistent state management with versioned persistence, and robust resource handling for MCP lifecycle and grammar-safe tool calling. Some duplication exists between two llama.cpp provider plugins, but overall architecture is well-structured.
SGLang is a high-performance inference serving system for large language models (LLMs) and multimodal models (vision, video, audio, diffusion). It provides an OpenAI-compatible API, a Python frontend DSL for programmable inference, and supports a wide range of hardware backends (NVIDIA CUDA, AMD ROCm, Intel XPU, Ascend NPU, Apple MLX). The architecture is a distributed serving runtime (SRT) with a scheduler, model executor, custom CUDA/Triton kernels, and advanced features like speculative decoding (EAGLE, MTP, NGRAM), disaggregated prefill/decode, hierarchical KV cache (HiCache), and quantization (FP8, NVFP4, AWQ, GPTQ). It also includes a Rust-based model gateway for routing and a multimodal generation pipeline for diffusion models. The system is designed for production-scale deployment, powering trillions of tokens daily across hundreds of thousands of GPUs.