Every AI model, cataloged.
LLM Reference helps developers choose or re-check a model + provider. We bring specs, pricing, benchmarks, availability, and official-source freshness into one decision-support reference.
What We Track
1,942
1,838 active today; deprecated public rows included
565
Grouped by architecture and lineage
147
Cloud platforms, API services, and inference endpoints
259
Labs and organizations building the models
168
Standardized evaluations for comparing model performance
- Pricing
- Context windows
- Parameter counts
- Release dates
- Knowledge cutoffs
Behind the Scenes
LLM Reference is built by Data Advantage, LLC — and maintained by research agents that audit catalogs, verify entries against official documentation, research papers, model cards, provider APIs, pricing pages, and announcements, and flag stale data. This provenance-first workflow keeps the reference current without presenting unverified user submissions as fact.
How the Data Works
- Static seed data — rebuilt at deploy time. No stale caches, no runtime dependencies.
- No user-generated content — every entry is researched and verified by our team.
- Official sources — specs come from research papers, model cards, and API documentation.
- Direct pricing — sourced from provider pricing pages, not third-party aggregators.
Built by Data Advantage
- Owner
- Data Advantage, LLC
- Purpose
- A public developer reference for evaluating models and providers.