Run 284B DeepSeek-V4-Flash (<59GiB RAM: 41.1 cold pp, 9.1 N=1 tg, up to 25.3 spec tok/s) & 30B MoEs (22.4 N=1 / 111.6 spec tok/s) on 64GB CPU (0 GPUs).
Repository Index
78,850 visible repositories in the index
73–93 of 93 · Most stars
🚀 A real-time monitoring dashboard for Claude Code & Codex, built with SQLite3, Node.js, Express, React, Vite, TailwindCSS, & WebSockets. It tracks sessions, agent activity, tool usage, and subagent orchestration, providing live analytics, a Kanban status board, status notifications, a cute buddy, & an interactive web UI/MacOS/Windows native app.
- claude-agents
- claude-code
- claude-skills
- express
Run models larger than your RAM or VRAM by streaming MoE experts from SSD. Built on llama.cpp.
Run large AI models on your own PC: Qwen3.8-Flash-Next on Strata's engine, and GLM 5.3 Flash, DeepSeek V4 Flash, LFM2.5, Qwen3.6, Ornith 1.5, Gemma 4, Tiel Coder and MiniCPM-V on a second engine written for PolyStrata. One-click install for Windows and Linux, OpenAI/Anthropic API on localhost.
- anthropic-api
- cuda
- gemma
- gguf
Expose Gemini CLI endpoints as OpenAI API with Cloudflare Workers
A code repository indexing tool to supercharge your LLM experience.
- neovim-plugin
- retrieval-augmented
- embeddings
- rag
37 tok/s from Qwen3-Next-80B (48 GB) on a 32 GB laptop, 24 without drafts vs 2.8 for llama.cpp; reading 62 vs 16 tok/s. Faster than llama.cpp on all 13 models compared in Q4_K_M. LLMs bigger than your RAM, pure C, CPU or GPU.
- avx2
- c
- cpu-inference
- gguf
AI drew your research figure. Tavotto makes it publication-ready: edit matplotlib and AI-generated figures on a canvas, lay out multi-panel pages in mm, check journal rules before export. Scripts stay untouched.
- academic-publishing
- ai-tools
- claude-code
- codex
Run a 105 GB AI model on a Mac that can't hold it. Qwen3.8-Flash-Next 4-bit, streamed from SSD on 16–64 GB Apple Silicon Macs. Fully native Swift, MLX and Metal. Offline, no Python. CLI, Swift library, and Ollama/OpenAI/Anthropic APIs. Works with Claude Code and Codex.
- apple-silicon
- llm
- llm-inference
- local-llm
Xolotl: run DeepSeek-V4.1-Flash, all 552B, as a local coding agent on one gaming PC (RTX 4090 + 110 GB RAM): ~33 tok/s on agent tasks, ~41 in fast mode, 512K context. Plus GLM-5.3-Flash, Qwen3.8 and LFM3-Omni. MoE experts tiered across VRAM, pinned RAM and NVMe, with Xolotl Direct NVMe->VRAM. OpenAI + Anthropic API.
Build AI text detectors whose decisions are mechanistically grounded and human-auditable (EMNLP 2026)
- ai-text-detection
- explainable-ai
- interpretability
- mechanistic-interpretability
🧱 Experimental library for building inference engines
UCE is a zero-shot foundation model for single-cell gene expression data
Can AI agents rebuild websites through browser exploration?
- benchmark
- harbor
- website-reconstruction
- agent-evaluation
C++/CUDA single-GPU inference engine for Qwen3.8 (incl. Qwen3.8-Flash-Next with offloaded experts) on the RTX 5090; grew from NInfer
A ChatGPT, Gemini, Claude, and Ollama(offline) plug-in for Embarcadero RAD Studio IDE XE5 to 12.1
- chatgpt
- chatgpt-api
- delphi
- embarcadero
Pure-Rust inference engine for Qwen GGUF models (Qwen3 to Qwen3.8, dense and MoE up to 770B+). CUDA and Vulkan (NVIDIA, AMD), hybrid CPU+GPU experts, MTP speculative decoding, OpenAI and Anthropic compatible server. Up to ~1.7x llama.cpp decode on a 12 GB RTX 3060.
PROMETHEVS: run the full DeepSeek-V4.1-Flash-1017 (510 GB) on a gaming PC with 64 GB RAM and a 12 GB GPU. PROMETHEVS engine (compress-then-correct), OpenAI-compatible local server, report and logs.
A browser bookmarklet for exporting conversations with ChatGPT as markdown files.
- bookmarklet
- browser-bookmarklets
- browser-tools
- chatgpt
Research experiments exploring uncommon quant techniques.
- analysis
- jupyter-notebook
- python
- quant
Lightweight Go SDK for Yandex Messenger Bot API — type-safe client with retry, rate-limit handling, and service-oriented architecture
- api-client
- bot
- bot-api
- chatbot