Week of Oct 26 to Nov 1, 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, November 2.
Text-to-3D model by OpenAI that generates textured meshes and NeRFs.
Python agent library that lets an LLM operate a real Chrome browser to finish web tasks.
Generates smooth in-between animation frames from two cartoon keyframes using video diffusion.
Open-source AI code assistant for VS Code and JetBrains connecting to any LLM.
NVIDIA's multi-framework model server with dynamic batching, ensembles and HTTP and gRPC endpoints.
Application framework that brings Spring design principles to AI engineering for Java teams.
All-in-one video model unifying reference-to-video, video-to-video, and masked editing.
CLI and library for testing, evaluating, and red-teaming LLM outputs.
Once the standard open TTS toolkit, now unmaintained; the Idiap fork on PyPI as coqui-tts carries it on.
Foundation model by Meta for promptable image and video segmentation.
Weak supervision system that generates training labels programmatically from labeling functions.
Data processing system for foundation models with over 200 composable operators for multimodal data.
Web UI for FLUX LoRA training on GPUs with 12 to 20 GB of VRAM, built on Kohya sd-scripts.
Pure and Lightning ID customization for diffusion models without fine-tuning.
Latest Llama model family by Meta with Mixture-of-Experts architecture.
Lightweight, scalable Python LLM inference and serving framework focused on high throughput.
RL framework that trains multi-step LLM agents with GRPO behind an OpenAI-compatible client.
Open music foundation model family covering song generation, a 12.5 Hz codec, transcription, and alignment.
Language detection library covering 75 languages that stays accurate on single words and short phrases.
Neural OCR model by Meta for academic documents and mathematical expressions.
Self-hosted RAG engine that focuses on layout-aware document parsing, editable chunks and cited answers.
Offline speech recognition toolkit supporting 20+ languages with small models.
End-to-end TTS model combining a conditional VAE, adversarial training, and normalizing flows.
Self-hostable chat workspace with multi-provider support, plugins, knowledge bases, and agents.
PostgreSQL extension adding DiskANN-style vector indexing and quantization on top of pgvector.
NVIDIA's open world foundation models that generate physics-aware video for robotics and autonomy.
Realtime voice conversion client for RVC, Beatrice, so-vits-svc, and DDSP-SVC models.