TheClawBay provider extension for Pi Coding Agent - access GPT-5, Codex, Claude, Gemini, DeepSeek, open-weight, and image models

AI SDK (Vercel) provider for Mixlayer — open-weight text and vision model inference over an OpenAI-compatible endpoint.

Jacobian-Lens interpretability primitive for open-weight LLMs (Anthropic 2026-07-06, 'Verbalizable Representations Form a Global Workspace'). Reads the model's functional workspace at inference time — lens_l(h)=unembed(J_l·h) — into workspace tokens, laye

The **GMI Cloud provider** for the [AI SDK](https://ai-sdk.dev/docs) contains language model support for [GMI Cloud](https://www.gmicloud.ai), offering GPU inference for open-weight models over an OpenAI-compatible API.

Local privacy router for coding tools: pseudonymises every request before it leaves your machine, and sends work a frontier model would refuse to an uncensored open-weight model you control.

A coding-agent CLI that works the same way whether the model is Claude, GPT, or an open-weight model like Qwen, DeepSeek, GLM, or Llama running locally.

TypeScript SDK and CLI for Open Security, a fork of Codex Security for open-weight models

Install and run HelpOfAi, the agentic terminal for open-source and open-weight coding models, from GitHub release artifacts.

Private vibe-coding LLM in your own AWS account — open-weight models on 100% spot instances with placement-score intelligence

Run open-weight models on your own Modal account from inside pi, shared across a team

Pi extension for the Hetzner Experiments Inference API — OpenAI-compatible open-weight models, catalog aligned with models.dev

Deterministic calculation toolkit for AI agents, especially local/open-weight models (Llama, Qwen, Mistral, etc.) that reliably fail at exact math, dates, and combinatorics: exact arithmetic, date math, unit conversion, statistics, subnet math, and more.

Minify font declarations with PostCSS

Scan-derived runtime guard for open-weight models — screen text against the exact glitch tokens Ingot's deep battery measured corrupting on your checkpoint.

Run and fine-tune Inkling-Small, Inkling, and other open-weight models with Tinker from Pi: effort sweeps, evals, SFT scaffolding, checkpoint chat, and Cookbook export/serving plans.

Set up a planner → local executor → auditor AI coding pipeline: installs Ollama, picks a right-sized open-weight model for your hardware, and installs the skill into Claude Code, Codex CLI, Cursor, and Windsurf.

Self-host the Open Sans font in a neatly bundled NPM package.

Hetzner Experiments Platform Inference API for the pi coding agent: free open-weight models, live catalog discovery and rate-limit visibility

Merge CSS rules with PostCSS.

Simplified Chinese (GB2312) fallback fonts for EmbedPDF - Noto Sans Hans (5 weights)