High-performance, AI-native server built from scratch in C + hand-written Assembly (SSE2/AVX2/AVX-512). Supports 10 AI providers — OpenAI, Groq, Anthropic Claude, Google Gemini, DeepSeek, Moonshot/Kimi, Zhipu GLM, Perplexity, Mistral AI, and local models via Ollama — all in one binary.
About this project
⚡ NeuroHTTP — AI-Native Web Server (C + ASM) High-performance, AI-native server built from scratch in C + hand-written Assembly (SSE2/AVX2/AVX-512). Supports 10 AI providers — OpenAI, Groq, Anthropic Claude, Google Gemini, DeepSeek, Moonshot/Kimi, Zhipu GLM, Perplexity, Mistral AI, and local models via Ollama — all in one binary. 📖 Arabic Version — النسخة العربية 📋 Table of Contents Features Supported Providers Prerequisites Build & Run Configuration Adding API Keys Adding AI Models API Endpoints Plugin System Makefile Reference Project Architecture Benchmarks Troubleshooting 🔥 Features ⚡ Performance Core Event-driven engine — single-threaded epoll edge-triggered event loop with iouring detection iouring async I/O — full iouring integration for zero-syscall I/O; falls back to epoll gracefully; SQPOLL support (Linux 5.11+) Hand-written ASM — CRC32 (SSE4.2), JSON tokenizer (SSE2/AVX2), memory copy (SSE2–AVX-512), non-temporal stores Zero-copy I/O — sendfile/splice support for efficient static file serving Memory arena allocator — bump allocation + slab allocator for lock-free per-request memory
From the project README on
GitHub
- Stars
- 53
- Forks
- 3
- License
- other
- Last push
- 29 May 2026
Add this badge to your README
Show that your project is listed on Made in Algeria.
[](https://www.madeinalgeria.dev/projects/neurohttp)