お気に入り

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.

小説や脚本から完成映像へ。セルフホストできる AI 動画ワークベンチ。

汎用 AIGC 動画エンジン：脚本から完成作品までを一つのパイプラインで。

文字起こし・字幕・LLM によるクリップ抽出をローカルで。

手間ゼロの動画編集：無駄な時間を自動でカット。

好きな動画を見せると、同じスタイルの新しい動画を作る。

開発者のための決定論的アバターとイラスト — 30+ スタイル、1 つの API。

無料でオープンなピクセルアート多機能ツール — スプライト・アニメ・パレット。

万能 2D エディタ：ピクセル・ベクター・アニメ・ノードグラフを一つに。

Claude を任意のゲームに向ければ、ほぼ何でも MOD できる。

あらゆるエージェントから、実戦品質の画像・動画・音声・3D を生成。

画像を渡すとゲーム対応の 3D アセットに — Apple Silicon 上で。

AI が仕切る、あなた専用の一人ゲームスタジオ。

一般向け GPU で動くローカル楽曲生成

Open music model with symbolic planning and zero-shot covers

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

曲をボーカル・ドラム・ベース等に分離

Training and inference toolkit for Stable Audio open models

MusicGen and companion models for text-to-music

Diffusion-based singing voice synthesis toolkit

ACE-Step 1.5 XL ベースの Windows ワンクリック音楽スタジオ

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