# LLMap

> The navigable archive of AI inference: 376 models, 71 devices (GPUs, Macs, phones),
> 79 inference providers with 1544 live price offers — cross-linked,
> synced from primary sources daily, with provenance on every fact.

## Data

- [Models catalog](https://llmap.ai/models): specs, quantizations, VRAM requirements, licenses
- [Hardware catalog](https://llmap.ai/hardware): GPUs and Apple Silicon with fit verdicts per model
- [Providers](https://llmap.ai/providers): hosted inference pricing, privacy and data-residency terms
- [Benchmarks](https://llmap.ai/benchmarks): leaderboards with provenance, score-per-dollar rankings
- [News](https://llmap.ai/news): model releases and price changes, detected within 24 hours
- [Licenses](https://llmap.ai/licenses): parsed license terms for every model

## Insights (open data, CC BY 4.0)

- [Insights hub](https://llmap.ai/insights): daily, citable charts on the AI inference market
- [LIPI](https://llmap.ai/insights/lipi): LLMap Intelligence Price Index — the price of intelligence, daily (CSV: https://llmap.ai/api/v1/insights/lipi.csv)
- [Attention share](https://llmap.ai/insights/market-share): open vs proprietary and US vs China share of OpenRouter tokens (CSV: https://llmap.ai/api/v1/insights/attention.csv)
- [Competitor map](https://llmap.ai/insights/competitor-map): intelligence vs price scatter of every scored model (CSV: https://llmap.ai/api/v1/insights/competitor-map.csv)

## API

- REST: https://llmap.ai/api/v1 (models, hardware, providers, benchmarks, search)
- MCP server: https://llmap.ai/mcp (streamable HTTP — use LLMap as a data source in any agent)
- Open Model Cards: https://llmap.ai/api/v1/models/{slug}/omc.json (openmodelcard.org spec)

Every API result includes source_url and data_synced_at.