MiniMax-H3 33B audio+video DiT accelerated 1.54x on a single NVIDIA DGX Spark (GB10, sm_121): NVIDIA Sol-Engine FirstBlockCache + batched VAE decode ported to single-GPU ComfyUI, Sol-Attn patched for compute 12.1, kijai Triton INT8/TMA kernels. Benchmarks, charts, one-shot installer, agent recipe.
A local web cockpit that runs several AI coding agents side by side; your keys, your machine.
AI gateway patterns for a compatible interface, model routing, proxy deployment, virtual keys, spend tracking, and Argolink integration.
Lighter airdrop tracker β track Lighter points, trading volume and leaderboard rank in the terminal, tick season farming tasks and estimate your airdrop allocation and eligibility. Read-only: public Lighter API, no private keys. Unofficial community tool, not affiliated with Lighter (lighter.xyz).
Android attestation bypass toolkit (WIP, contains vibeslop)
Drive the OpenAI Codex Micro macro pad from Claude Code. Six Agent Keys show which of your projects needs you, and pressing one jumps to its terminal tab. Every key, the dial and the thumbstick, over USB or Bluetooth. No ChatGPT desktop app required. Includes the fully documented device protocol.
GLM-5.3-Flash (Z.ai) NVFP4 (LibertAI) + NVFP4 KV, 1M context on 2x DGX Spark GB10 β vLLM TP2 recipe, walls ledger, benchmarks, needle to 1.04M; abliteration with blackforest_AI next
MiniMax H3 (video + audio) on TensorFold 0.6.5 with an H3 family, Mac Studio M5 Ultra: 5 s 864x480 clip in ~50 s, 40 s at 768x448 β the fastest H3 we have measured on Apple Silicon; int8 tensor-unit kernels + Turbo adapter; one-shot + GHCR carrier
Graphical macro builder for Linux Mint. Chain steps: open apps, files, sites, commands; place windows; click; press keys; type text. Save them as one-click Desktop/menu shortcuts. Like AutoHotkey, on top of xdotool/wmctrl. One self-installing Bash script, ES/EN interface, GPLv3.
Local desktop app to build structured MiniMax H3 video prompts (T2VA/I2VA/FL2VA/L2VA/Ref2VA) using a local LLM via Ollama or LM Studio. Streams output, auto-describes reference images with vision models, 100% offline β no API keys, no cloud.