0xSero

TurboQuant: Near-optimal KV cache quantization for LLM inference (3-bit keys, 2-bit values) with Triton kernels + vLLM integration

Your AI friend right in your browser

Dot-lattice art: photo to two-tone dot map to seamless animated loop (engine + scripts)

Model-agnostic MoE compression automation: build calibration bundles, run REAP/quantization/benchmark/publish stages, and render auditable reports.

Ship your repo + live coding-agent session (Claude Code / Codex / pi / Droid) to another machine over Tailscale; it resumes in tmux and keeps working.

Local AI for Omarchy: the model validated for your GPU, one button on the bar. Start serves it, any coding agent opens on it, one click shares it on your tailnet.

Pinned one-DGX-Spark Docker recipe for DeepSeek V4 Flash with EXL3, SparkInfer, and 262K NVFP4 MLA KV cache

Browse the world in the comfort of your terminal

GLM-5.2-NVFP4-REAP-469B serving on SM120 (4× RTX PRO 6000 Blackwell) — one-command vLLM launch recipe, 250K context, DeepSeek Sparse Attention + MTP speculative decode

How much experts do we need to serve a model?

DeepSeek-V4.1-Flash on two DGX Sparks: EXL3 routed experts, 262k context, 2M-token KV, vision, tools, MTP, bundled with Pi

Point any coding harness (Claude Code, Codex, OpenCode, Pi, OMP, Crush, Copilot, Grok…) at any OpenAI/Anthropic/Responses-compatible inference endpoint. Core library + CLI + macOS tray + web UI.

Anthropic-compatible HTTP facade over claude-agent-acp

Unified BYOK proxy that wires ~/.factory/config.json models into Cursor IDE

Organise your AI's memories with graph database entries

An Opencode plugin for managing git worktrees.

REAP expert pruning for MoE LLMs on Apple Silicon via MLX