Week of Sep 28 to Oct 4, 2026
Tools of the Week
Every Monday this page reshuffles and puts a different tool forward from each of the 27 categories in the directory. It is a way to surface the good work that never reaches the top of a ratings list. 27 picks are live right now, and the next rotation lands Monday, October 5.
Generates high-fidelity 3D meshes from single images with a 1.5B rectified flow transformer.
Terminal coding agent that runs commands in native sandboxes, optimized for low-cost models.
Animates a still human photo with 3D SMPL parametric motion guidance extracted from a driving video.
Self-hosted alternative to GitHub Copilot using open-source models.
Open-source orchestrator for AI training and inference across clouds, Kubernetes, and bare metal.
Gradient boosting library by Yandex with native categorical feature handling.
Feature-rich web UI for Stable Diffusion with extensive extension ecosystem.
Meta's layered guardrail framework that scans agent prompts, plans, and generated code.
Whisper with word-level timestamps and speaker diarization
Open-vocabulary object detection model by Google using vision transformers.
Simple image annotation tool for creating YOLO format labels.
Detects label errors, outliers, and duplicates in machine learning datasets using confident learning.
Graphical interface for kohya-ss SD training scripts.
Zero-shot identity-preserving image generation from a single face photo.
Ai2's fully open post-training recipe, datasets, and code for building state-of-the-art instruct models.
High-throughput LLM serving engine with PagedAttention
Low-code framework for building custom AI models by Predibase.
Open toolkit from Alibaba for text-to-music generation with 48 kHz output and multi-minute long-form pieces.
Library for efficient text classification and word representation by Meta.
Python library for extracting text, tables, and metadata from PDFs.
Python stream processing framework with a Rust engine for real-time analytics and live RAG pipelines.
Edge-optimized speech recognition models that beat Whisper on speed and accuracy for on-device use.
Instant voice cloning TTS by MyShell requiring only a short audio reference.
SvelteKit chat frontend for OpenAI-compatible APIs that powers Hugging Face's HuggingChat.
High-performance embedding server by Hugging Face for production deployment.
Unified post-training and inference framework that accelerates open video generation models.
Advanced RVC fork with enhanced training and inference features.