Your favorites

Record everything on your screen locally and let AI search and act on it.

Point it at a question; it browses the web and hands back a sourced report.

Ask your database questions in plain English; it clarifies, then writes the SQL.

Give your agent a disposable Linux sandbox to run untrusted code safely.

Turbocharge LLM/VLM serving with 4-bit weights and a fast gRPC API.

Turn messy PDFs and Office docs into clean Markdown, JSON or a layout tree.

Ask questions across your own notes, files and chats from a chat box.

Embed, index, search and prompt — no separate vector DB to run.

One import to talk to OpenAI, Anthropic, Google, Mistral and 100+ more.

Pull tables out of PDFs into DataFrame with almost no setup.

The canonical Skill library — copy any SKILL.md and your agent gains a capability.

Drop-in skills that make your coding agent actually do TDD and debug properly.

Point it at a docs site and get a ready-to-use SKILL.md back.

Make your agent stop shipping generic 'AI slop' front-ends.

Give your agent a durable, file-backed plan it actually follows.

Domain skills that let an agent do real scientific analysis.

Make AI drafts read like a human wrote them.

The discovery index for finding your next agent skill.

Studio-quality TTS from an 82M model you can run on a CPU.

A vision-language model that fits on a laptop and reads video.

A 1.7B model good enough to be your default on-device LLM.

A 3.8B model that reasons better than models twice its size.

A 3B all-rounder with unexpectedly good coding and 29+ languages.

A 1B model that reads images and speaks 140+ languages with a 128K context.

The 22M embedding model behind half the semantic search on the planet.

A 3B model that sees, hears and reads — with excellent OCR.

Turn a novel or screenplay into a finished video, on your own server.

A general-purpose AIGC video engine: script to finished film in one pipeline.

Transcribe, subtitle and LLM-clip long videos — locally.

Effort-free video editing: it cuts the dead air for you.

Show it a video you love; get a new video in the same style.

Deterministic avatars and illustrations for developers — 30+ styles, one API.

A free, open pixel art multitool — sprites, animation and palettes.

One universal 2D editor: pixels, vectors, animation and node graphs.

Point Claude at any game — and let it mod almost anything.

Production-ready images, video, audio and 3D — generated from any agent.

Drop an image, get a game-ready 3D asset — on Apple Silicon.

Your own one-person game studio, orchestrated by AI.

Local full-song music generation that fits on a consumer GPU

Open music model with symbolic planning and zero-shot covers

Earlier ACE-Step stack — still useful if you follow v1 tutorials

Split any track into vocals, drums, bass and other stems

Training and inference toolkit for Stable Audio open models

MusicGen and companion models for text-to-music

Diffusion-based singing voice synthesis toolkit

One-click Windows studio built on ACE-Step 1.5 XL

Fast zero-shot TTS with natural prosody

Multilingual open TTS with an active community

Instant voice cloning with style control

The default WebUI for voice conversion covers

Word-level timestamps on top of Whisper-class ASR

Talking-head video from one photo plus audio

Practical super-resolution for photos, anime and frames

Keep the same face across generated images without heavy fine-tuning

Full songs with lyric alignment via block flow matching

Open lyric+tag music generation stack for 2026

Fully open long-form song model with style controls

Text-queried universal audio separation

Latent diffusion for text-to-audio and music

Desktop GUI for high-quality vocal/stem removal

Real-time AI voice changer for streams and games

Classic open singing voice conversion stack

Pixel-art agent that paints sprites tool-by-tool

Generate and export game-ready pixel sprite sheets

Sprite editor with AI assist and Godot export

Structured image-to-3D for game props

Fast feed-forward image-to-3D

Efficient image-to-3D meshes

Classic open text-to-3D playground

LLM-composed canvases for structured concept art

PuLID (Pure and Lightning ID customization) is an identity-consistent image generation method: one reference photo is enough to place the same character into new scenes, styles and outfits — without the heavy fine-tuning LoRA training usually requires.

Controllable camera/object motion for AI video

Image-to-video for animating concept art