#gpu

An agent harness that compiles a model into one provably-correct, self-retargeting CUDA megakernel and self-tunes it past cuBLAS at batch-1 LLM decode, paper: https://arxiv.org/abs/2606.09682

Run the Google Colab GPU MCP server on Termux/Android: source-built pydantic-core + maturin, DNS-over-HTTPS fix, one-command installer. Call Colab T4/L4 from opencode, Claude Code, Cursor.

Predictive GPU memory framework for AI inference workflows

Agent Skills and an MCP server for GPU performance profiling, benchmarking, optimization, and reporting, with an inference focus.

Connect local models, AI APIs and private machines into one controlled pool for apps and agents. Local-first, self-hosted AI orchestration.

Production-grade low-level engineering skills for AI coding agents: 193 verified, source-backed skills (C/C++/Rust/asm/kernel/networking/embedded/Zig/GPU, safety, PQC, eBPF, io_uring, secure boot, RE, design) — VFS fops, writeback, workqueues, timers, kthreads, sk_buff, TCP CC. Every claim traced: claim -> source -> section -> skill.

Private AI Circuit free tier. Run a private AI agent on AWS GPU. Hermes + Ollama + Terraform

Private AI platform on k3s — Ollama/Qwen3 + Milvus RAG + MCP agentic tooling + Fleet multi-cluster management

Agent-friendly CLI, MCP server and SDK for managing AutoDL GPU instances — create, boot, SSH, run and auto-shutdown, all scriptable. 用命令行和 MCP 管理 AutoDL GPU 实例。

Run open LLMs on free Colab/Kaggle GPUs and call them from your machine over an OpenAI-compatible endpoint

Local inference server in pure Mojo. Speaks OpenAI, Anthropic, and MCP. Models are OCI artifacts. Same source runs on CPU, NVIDIA, AMD, and Apple GPUs. Apache-2.0 all the way down.

A local-only AI agent for small devices

The 48GB VRAM Local LLM Guide: real benchmarks, model configs, and gotchas from 135+ community reports. Includes runnable benchmark scripts.

A cross-platform tool to discover and compare local LLMs, detect hardware specifications, visualize VRAM and RAM usage, and find the best models your device can run.

Fast local search over your text and Markdown files, built for AI agents: keyword (BM25), GPU semantic search and reranked hybrid retrieval, served over MCP.

AI-апскейл прокси-узел для AniLabX -> XRemoteDesktop: перехватывает воспроизведение и апскейлит поток в реальном времени через GPU-шейдеры

One command runs the best local LLM your GPU can handle: model picked from measured multilingual benchmarks, llama.cpp tuned for AMD, NVIDIA, Intel and Apple GPUs.