"Large Language Models"

Open MoE coding-agent checkpoint with 39.8B total and about 2.5B active parameters; BF16 weights under Apache 2.0, served with SGLang

can interpret and execute functions/plugins

a model designed for both function calling and on-device inference

Unified schema-based information-extraction model library supporting named-entity recognition, text classification, relation extraction, span attributes, and structured JSON extraction

Qwen3-0.6B decision model with LoRA, a scalar scoring head, shared state prefill and isolated candidate branches.

One-pass typed-decision model family fine-tuned from Qwen3.5, supporting Choice, Score, and Noul questions with calibrated probabilities

ModernBERT/mmBERT encoders with typed decision heads for choice, ordinal and boolean scoring; batches full question sequences, with English, multilingual and task-specific checkpoints

Open-model implementation of the Jev interface pattern supporting CUDA and MLX with typed option probabilities

Small Jev replica with dynamic candidates, shared decision heads, and end-to-end training

Qwen3.5-9B LoRA decision model that scores permitted answer tokens for choices, booleans and rubric levels, with MLX and PyTorch reference tooling.

Open-source semantic decision engine with Choice, Noul, and Score primitives, a ModernBERT zero-shot backend, and a TypeSafe-compatible /v1/systemone API

Small Qwen3.5 and Gemma 4 fine-tunes for zero-shot Choice, Noul, and Score classification, with an optional MLX server for Qwen models

Qwen2.5-0.5B decision model with LoRA, isolated packed question branches and a pointer head for typed option probabilities.

GLiFormer-based classifier with a Jev-compatible API for choices, boolean probabilities and rubric scores.

Independent starter model for one-pass scoring over variable-length text options, with CPU, Apple MPS, and CUDA support

Qwen3.5 NLI cross-encoder for entailment, contradiction, and neutral classification, with reranking and grading APIs and optional image support

Qwen3.5-0.8B typed-decision checkpoint for schema-bound scoring of text and image candidates, evaluated by its authors at 93.23% accuracy on 12,000 decisions

a multi-modal model from Apple

2-bit quantized micro language model that runs on Z80/CP/M (.COM), trained in Python and exported

pretraining of a 1.1B Llama model on 3 trillion tokens.

a tool that enables pair programming with LLMs to edit code in local git repositories.

a self-hosted AI coding assistant

Proxy that allows you to use ollama as a copilot like Github copilot

a prototype development tool that leverages GPT

a VSCode extension that uses llm-ls

an Emacs extension for using a local LLM

An open-source alternative to GitHub copilot that runs locally.

a MITM proxy that lets you use the GitHub Copilot extension with other LLMs

macOS computer-use agent that combines OCR with TypeSafe decisions and requires Screen Recording and Accessibility permissions

Goal-driven browser automation using Jev decisions over observed page elements instead of generated selectors or scripts

Standalone mobile agent using Jev decisions, with a CLI, React Studio, traces, and latency measurements

Adds screenshot, mouse and keyboard tools to Pi on macOS via an extension, skill and native helper.

An Electron app for macOS

Discord moderation bot using TypeSafe AI and Jev to filter spam and scam links, escalate offenses, and profile members

an on-device real-time multimodal AI app for local voice and vision conversations, powered by Gemma 4 E2B and Kokoro.

an intriguing personal assistant based on local data

Pi guardrails for blocking irreversible or off-task calls and detecting stuck loops

a terminal assistant that leverages tmux

Voice-controlled headed Chromium browser using Web Speech API, Playwright, and typed Jev intent and target decisions

an open-source LLM app development platform with a node-based UX

a TUI-based code assistant written in Go.

a toolkit for spec-driven development

a node-based GUI for quick iteration of langchain flows

a JavaScript environment for prompt development and structured data extraction for LLMs.

Natural Language Programming against multiple LLMs

a local language server that leverages LLMs

turn Python functions into LLM-powered endpoints using the OpenAI SDK

a lightweight NLP toolkit for tasks like fuzzy-deduplication, similarity, and ranking

a tool for compressing prompts with minimal loss of information

a tool for creating and managing LLM workflows

an open-source platform for evaluating, testing, and monitoring LLM applications.

Research collection for typed decision scoring, with local Qwen text inference and task-specific chess, Doom, memory and robot-control experiments.

a unified interface for data and feedback collection, including model comparisons

self-hostable toools for evaluating LLMs, vector databases, and prompts

a visual programming environment for benchmarking prompts across multiple LLMs

A tool for testing and evaluating LLM prompt quality.

A local, drop-in replacement for the OpenAI API

an OpenAI gateway in Go to create API keys with rate limits, cost limits and TTLs

Cloudflare Worker text-classification API backed by TypeSafe Jev, with batch requests and optional reasoning-model escalation.

a multi-node AI orchestration platform for deploying tool-using agents across server clusters, with CLI/browser interfaces, governance, and MCP support.

An automated installer for LLaMA

an intriguing P2P clustering solution for running models across several machines

a self-hosted inference engine for edge computing with an OpenAI API

A local (edge) inference engine in C++ without any dependencies

a tool that enables inference on AMD GPUs

a PyTorch extension for Intel hardware

iPad IDE that bundles an on-device LLM engine and a PCIDriverKit AMD GPU driver for local inference

A pure C/Objective-C+Metal inference engine that runs the 397B MoE model on Apple Silicon laptops using SSD streaming, 2-bit quantization, and custom GPU kernels.

Jev-compatible HTTP API server using open models and SGLang, with Modal deployment and autoscaling support

A high-performance llama.cpp fork with improved CPU and hybrid GPU/CPU inference plus modern quantization and DeepSeek/BitNet-focused updates.

Swift and Metal runtime that streams Qwen MoE experts from storage to run 35B and 80B models on Apple Silicon and iPhone.

C-based local LLM toolkit using llama.cpp, with CLI, HTTP server and MCP server; targets Linux x86-64 and Windows x86-64

Production Docker Compose stack and ROCm/vLLM patches for serving DeepSeek V4 Flash on a single AMD MI300X.

A C++ implementation of a GPT-2 inference engine

A C++ port of Facebook’s LLaMA model. Still requires roughly 240GB of (unoptimized) weights, but can run on a 64GB Mac.

A GPU-focused Python wrapper for LLaMa

A Rust port of llama.cpp

another C++ local inference tool

an alternative to Ollama for locally running models

a framework that allows users to serve thousands of fine-tuned models on a single GPU

a set of acceleration techniques

TypeScript SDK that detects, replaces, encrypts, and rehydrates PII in LLM prompts, coding agents, and browser workflows.

Python runtime and serving implementation for typed, calibrated decisions from text or image input, with a local TypeSafe SDK-compatible API

Jev-style decision API using permitted-answer token logits, shared prompt-prefix caching and batched question suffixes.

TypeScript/Bun Jev-compatible API bridge that translates decisions into classification prompts for a local DiffusionGemma model served through an OpenAI-compatible oMLX endpoint; probability estimates are not direct logits

MLX multimodal decision scorer with shared visual-context prefill, answer-label logits and teacher-forced candidate sequence scoring.

Jev-compatible decision server using DiffusionGemma answer-slot probabilities and repeated denoising reads through a patched vLLM backend.

Local bilingual decision API using frozen Qwen3-4B-Instruct answer-label logits, with an MLX backend and sequential question scoring.

Python wrapper for OpenAI-compatible clients that implements the Jev System One wire format and returns typed classification probabilities and confidence instead of prose

pure C implementation of Voxtral-4B-TTS for lightweight, dependency-free text-to-speech inference.

a plugin for Apple Notes that uses the Accessibility APIs

a rudimentary IRC bot that communicates with a local instance of ollama

a bot for integrating ChatGPT and DALL-E into Slack

Recipes for on-device voice AI and local LLM

Home Assistant integration for TypeSafe Jev with sensors, actions, and Assist conversation support

a CLI wrapper for ChatGPT

CLI to interact with LLMs using markdown and images

a terminal-based interface for LLMs

an easy-to-use AI text-generation software for GGML and GGUF models based on llama.cpp

a web-based interface for LLMs

a more or less sensibly designed self-hosted ChatGPT UI

A containerized solution for using local LLMs via web chat

a set of parsers for common file formats

a search SDK that unifies the entire RAG pipeline in a single database query

a tool for importing and chunking text files for RAG

a data pipeline designed to pre-process knowledge graphs and perform RAG on them

a local AI memory system that stores full conversation history in a structured searchable corpus with MCP tools and compressed context export.

a note taking tool that performs RAG using a local LLM

Offline email archive with local Gmail/IMAP sync, full-text search, DuckDB/Parquet analytics, and an MCP server for AI assistant query access.

AI-first knowledge base and startup OS

another framework to create bots from existing datasets

A way to quickly create custom embeddings off a web site

a framework for or rapid development and deployment of production-ready RAG systems with SQLite support

a framework for developing LLM-based applications including Retrieval Augmented Generation

a tool that researches a topic and generates a full-length report with citations

a Perplexity AI search engine clone

a local tool for searching using LLMs

A device that tells bedtime stories to kids, using chunked TTS

automatic generation of visualizations and infographics

A GPTScript tool to generate images

a set of tools for manipulating GGUF format files

a Python script for plotting the intermediate layer outputs of Mistral 7B

LLM training in simple, raw C/CUDA

an implementation of Half-Quadratic Quantization (HQQ)

a way to quantize the LLaMA weights to 4-bit precision

A way to do LoRA adaptation of LLaMa

Another way to do LoRA adaptation of LLaMa

Gemma 4 26B-A4B inference in ~2 GB of RAM on M-series Macs

Evaluation harness and published reports comparing Jev and GLiNER2.5 on intent and text classification, including latency and probability behaviour

Reproducible decision-model benchmark suites and corrected reports comparing Jev with LLM baselines, including accuracy, calibration, latency and choice limits

Benchmark and orchestrator for evaluating AI agents on network troubleshooting, with fault injection, MCP telemetry, sandboxing, and remote execution

Automatic language-model decensoring tool using directional ablation and Optuna-based optimisation to reduce refusals while limiting divergence from the original model

run large Mixture-of-Experts language models well on memory-constrained hardware

an implementation of word2vec skip-gram for word embedding.

A cross-platform dekstop application with chat history and extension support

an interesting take on how to use semantic search and OpenSearch’s BM25 implementation

an ensemble setup with WhisperSpeech, WhisperLive and Phi

The Modular Platform, including MAX and Mojo

a componentized approach to building LLM pipelines

a simple, lightweight LLM wrapper

Yet another alternative to langchain

yet another LLM framework using decorators for data validation

a langchain alternative with slighly better internal coding standards

Control modern language models more effectively and efficiently than traditional prompting or chaining.

a composable approach for building LLM applications

a data framework for LLM applications

has spinoffs for chat, workflows for medical/scientific papers, semantic search for developers and semantic search for headlines and story text

Yet another alternative to langchain, but with an interesting approach at defining workflows

an open-source ChatGPT alternative that runs 100% offline (uses nitro)

a self-hosted local inference front-end for chatting with document collections

another self-hosted local inference front-end

A VS Code notebook interface for LLMs

a Jupyter plugin that can handle code generation and image generation, but not switching models (GPT-4)

a Jupyter plugin that can handle code generation and fixes based on tracebacks

a Jupyter plugin that can handle multiple models

a Python package for serving large language models on Apple silicon.

a library that provides a response format for conversation structures and reasoning output in OpenAI models.

a lightweight library for efficient text chunking in RAG applications.

a Python library that uses LLMs to perform data cleaning

library for prompting, synthetic data generation, and training workflows

a vactorization library that can handle more than just text

decorators to create functions that return structured output from an LLM.

a package for validating and correcting the outputs of large language models

a memory management/summarization technique for unbounded context

a clever library that simplifies invoking OpenAI function calls

A simple wrapper for the ChatGPT AI

official AINA source for Catalan language models, embeddings, datasets, benchmarks, and demos

an example of how to do hyrid (vector and FTS) search with SQLite for RAG

tips on how to use OpenAI JSON and function calling

Manual instructions for Apple Silicon

a set of lecture notes and detailed examples of prompting techniques

a collection of LLM resources that operate independently

A set of sample base prompts for various LLMs

Another set of prompting techniques and detailed examples

might be a short-lived resource, but an interesting one

A minimal PyTorch implementation for training your own small LLM

a simple PyTorch-based implementation

A pretty hilarious image-to-description example

A self-hosted ChatGPT alternative

provides an example of how to use Graph navigation and Cognitive Search indexing

open-source LLM chatbots

An LLM-powered advanced RAG pipeline built from scratch

A simple interactive web shell with some nice features

a tutorial on how to build a GPT model from scratch

a collection of one-liners for various LLMs

porting of LLMs to tensor processors, starting with the Apple Neural Engine (ANE)

an embedding database

A simple vector database that can run in-process

A vector database that performs vector generation internally

A Single-File Vector Search Engine

a pretty complete GPT/search integration solution with GitHub, Slack and Confluence/JIRA connectors

Hugging Face model page for a MoEspresso V2 variant

A tutorial on LLM serving using MLX