The AI landscape,
made legible.
Find your way through open-source AI. Seventeen focused indexes connect the tools, models and infrastructure behind your next project.
60967 entries across dated index snapshots. Tools can appear in more than one index.
What are you building?
Choose a discipline. Find your starting point.
Build & connect
5 indexesExplore frameworks that give models tools, memory and room to act.
Explore coding-agent projects and their public repository signals.
Reusable instructions, commands and extensions for your next agent.
Discover servers listed in the official Model Context Protocol registry.
Check source coverage on the index; this count is not proof of a complete registry.
Explore gateways, routing and inference infrastructure for AI applications.
Models & media
5 indexesExplore public models, their categories and available source signals.
Explore the tools for running language models on your own hardware.
Find training, adaptation and optimization tools for your next model.
Explore the open-source tools behind generative visual workflows.
Speech recognition, synthesis and real-time voice tools, in one place.
Knowledge & data
4 indexesDiscover retrieval, vector search and knowledge-grounded AI tools.
Explore document parsing, extraction, OCR and knowledge workflows.
Explore tools to create, label, curate and manage machine-learning data.
Explore prompt engineering, reusable templates and optimization tools.
Measure & discover
3 indexesFind the tools to measure quality, reliability and risk.
Tracing, monitoring and debugging tools for AI systems.
Explore developer repositories and the public activity around them.
No indexes match. Try a broader search.
Find the index for the decision you are making
Start with your task, then open the relevant catalog and its methodology. The fleet combines different sources and scoring methods; counts can overlap and are not a count of unique products or independently tested recommendations.
Guide sources reviewed . Catalog snapshot: . Source review does not mean runtime testing.
Build an application
Explore agent frameworks, prompts, RAG, MCP connectors, document processing and voice components. Follow a project to its own source and verify how it fits the complete workflow.
Run or adapt a model
Use Model Radar for model discovery, Local LLM for runtime layers, Fine-Tuning for adaptation tooling and Diffusion for image/video workflows. Verify code and model artifacts separately.
Measure and operate
Use Eval for evaluation approaches, Observability for inspecting runs, Gateway for request routing and the shipping trackers for public activity signals. Define success before comparing candidates.
Evidence to check before choosing
- Choose a component role before comparing popularity or activity scores.
- Check source dates, scope notes, licenses, versions and known limitations.
- Use a representative, permitted task trial to evaluate suitability; no cross-index score is a universal quality ranking.
What remains unknown
The hub does not independently benchmark all indexed projects, deduplicate all overlapping tools, or verify production suitability and AI citation outcomes.
Sources and scope
Continue by task: agents · modelradar · eval.
About The Living Indexes
Snapshot: · 17 catalog records.
Source: Published sibling index snapshots. Collection and scoring source · Current JSON snapshot · Page inventory.
What does this catalog cover?
The Living Indexes contains 17 records in the snapshot dated 2026-11-12T17:38:39.645737+00:00. Coverage is defined by its published data source and collection rules; inclusion is not an endorsement.
How is the default order determined?
This hub reads each sibling data.json and links to its own methodology. GitHub-discovery directories, commit-velocity trackers, the official MCP registry directory and Hugging Face Model Radar use different signals. Counts sum catalog records rather than unique products across the fleet; unknown sibling counts are omitted from the sum.
When was this snapshot collected?
The recorded snapshot time is 2026-11-12T17:38:39.645737+00:00. The refresh workflow is scheduled daily; this timestamp records the available data rather than guaranteeing a successful refresh every day.
Browse The Living Indexes by topic. Published by Kymata Labs.