We check every product on this page once a day and record whether it still answers.
An open-source framework for running inference and evaluation on generative world-model AI (video, 3D/4D, embodied agents) via CLI/TUI/Studio interfaces — research infrastructure for AI developers, not an end-user app.
Open source
What to know▼
How it works
An open-source framework that unifies inference and evaluation for generative world-model AI, supporting video generation, interactive worlds, 3D/4D representations and embodied models through TUI, CLI and Studio interfaces, with a growing collection of integrated models and 58 benchmarks.
What's different
Unifies inference and evaluation for world models under one workflow instead of requiring separate tooling per model type.
Pricing
Open-source.
Best for
AI researchers and developers working on world models who need a common infrastructure for running and benchmarking them.
We check every product on this page once a day and record whether it still answers.
An open-source infinite canvas that combines notes, diagrams and an AI agent that writes results directly onto the board — built to replace juggling Notion, Excalidraw and a chat window.
Open sourceSelf-hosted
What to know▼
How it works
An open-source, real-time collaborative canvas. Notes, hand-drawn diagrams and interactive mini-apps live as nodes on one board; an AI agent reads the board, searches the web, and writes results back as new nodes instead of a chat reply.
What's different
Replaces switching between Notion for notes, Excalidraw for diagrams and a separate AI chat window — everything, including the agent's output, lands on the same canvas.
Best for
Teams and individuals who think visually and want an AI agent's output to live alongside their own notes and diagrams, not in a separate chat log.
We check every product on this page once a day and record whether it still answers.
An open-source Python backtesting engine that traces trading signals through weights, orders, fills, and performance for reproducible research, for quant developers, instead of building a custom backtester from scratch.
Works offlineOpen source
What to know▼
How it works
Supports vectorized research or order-aware simulation with transaction costs, walk-forward testing, 80+ metrics, and 25+ charts; runs locally, with optional managed market data through an API.
Pricing
Open-source and runs locally; managed market data available through a paid API.
We check every product on this page once a day and record whether it still answers.
An open-source 20-billion-parameter image generation model from Alibaba with strong Chinese and English text rendering — for developers and researchers building on top of it, not an end-user app.
FreeOpen source
What to know▼
How it works
An open-source 20-billion-parameter image generation model from Alibaba's Tongyi Qianwen team, with commercial-grade text rendering in both Chinese and English.
What's different
Leads in nine benchmark tests and offers strong bilingual (Chinese/English) text rendering in generated images, and is free and open-source.
Pricing
Free to use.
Best for
Developers and researchers building on top of an open-source image generation model rather than end users looking for a finished app.
We check every product on this page once a day and record whether it still answers.
A free, no-signup game where you try to jailbreak open AI agents guarding a secret behind a published system prompt, competing weekly for cash prizes.
FreeNo sign-upOpen source
What to know▼
How it works
Each challenge is a live, open-source AI agent guarding a secret with its system prompt published; you try to talk it past its own defenses, and landing the most approved breaks in a week wins rewards.
What's different
System prompts are published openly rather than hidden, and challenges rotate weekly.
Pricing
Free to play, no account needed.
Best for
People interested in AI red-teaming or prompt-injection who want to compete for cash prizes.
We check every product on this page once a day and record whether it still answers.
An open-source, self-hostable framework for running controlled experiments across your AI pipeline to statistically prove which changes actually improve output quality — for ML engineers, unrelated to restaurants.
Open sourceSelf-hosted
What to know▼
How it works
Treats every knob in an AI pipeline — retrieval, reranking, prompts, models, tools — as an experimental factor, runs factorial experiments, evaluates outputs with a configurable LLM or human reviewers, and applies mixed-effects models to measure which techniques help, by how much, and whether it's statistically significant.
What's different
Open source and self-hostable, replacing guesswork with statistically grounded experiment results.
Best for
ML engineers trying to prove which changes to an AI pipeline actually improve quality.
We check every product on this page once a day and record whether it still answers.
A visual mapping tool for linking people, places, vehicles and events on one canvas instead of scattered spreadsheets and screenshots - built for investigators and security analysts.
Open source
What to know▼
How it works
An infinite zoomable canvas where you create and connect entities — people, places, vehicles, events, organizations, communications — to visually map relationships instead of tracking them across scattered notes, spreadsheets, screenshots and documents.
What's different
Open source.
Best for
Investigators, researchers, journalists, compliance reviewers and OSINT analysts working with complex networks of relationships.
We check every product on this page once a day and record whether it still answers.
An open-source platform for building AI-led interviews over voice, chat, or video, self-hostable instead of relying on a closed managed interview tool.
Open sourceSelf-hosted
What to know▼
How it works
Design adaptive interviews, share a link, and the AI conducts the conversation over voice, chat, or video with contextual follow-ups and structured reports; self-host with Docker on a Next.js/Supabase/PostgreSQL stack with pluggable LLM providers.
What's different
MIT-licensed and fully self-hostable.
Best for
teams that want to self-host their interview data instead of using the managed service.
