AI Model Reports
AI Model Research Library
Market research on model performance, adoption, economics, and production choices.
Selected reports
8
articles
Families
10
Major model families
Industry sources
11
Source Landscape
Signal types
4
coverage areas
Source Landscape
External source library
External reports, benchmark sites, datasets, and official docs for AI model performance, adoption, cost, and availability.
2026 AI Index Report
Stanford HAI
Broad industry takeaways across technical performance, economy, science, policy, education, and public opinion.
Annual report cadence means fast model launches can move faster than the report cycle.
Trends in Artificial Intelligence
Epoch AI
Long-run trend rates for training compute, training cost, hardware, context windows, and inference prices.
Several estimates depend on frontier-model disclosures and confidence intervals rather than complete public reporting.
Models: Intelligence, Performance & Price
Artificial Analysis
Comparing models across quality, output speed, latency, context window, and price.
Benchmark coverage and pricing assumptions are third-party measurements, not provider-owned claims.
State of AI: 100T Token Study
OpenRouter + a16z
Real-world multi-model usage patterns across tasks, geography, retention, open vs. closed source, and cost dynamics.
The sample reflects OpenRouter traffic, so it is strongest for developer and multi-model usage patterns.
Anthropic Economic Index
Anthropic
Claude usage patterns, occupational exposure, geography, delegation, and economic impact signals.
It is a provider-specific window into Claude usage rather than the whole AI market.
State of Enterprise AI 2025
OpenAI
Enterprise ChatGPT adoption, reasoning-token growth, and product integration signals.
It is first-party OpenAI enterprise data, so it should not be generalized to all providers.
Signals Consumer Data
OpenAI
Consumer ChatGPT adoption, work vs. non-work use, region, and demographic views.
The dataset excludes enterprise and Codex usage, so it underrepresents business and technical use.
State of Generative AI in the Enterprise
Menlo Ventures
Enterprise AI spending, application-layer share, procurement patterns, and vendor category movement.
Survey and market model results should be treated as market estimates, not transaction-level records.
Top 100 Gen AI Consumer Apps
Andreessen Horowitz
Consumer AI app rankings, multi-tenant behavior, and category breakout patterns.
Rankings rely on third-party traffic and app intelligence sources rather than first-party product telemetry.
State of AI Report 2025
Air Street Capital
A narrative report spanning research, industry, and politics with practitioner commentary.
It is a report deck and synthesis layer, so primary data should be traced back when making numerical claims.
Arena Leaderboard
LMArena / Arena
Crowd preference rankings across model tasks, with vote counts, model scores, price, and context windows.
Crowd preference can be affected by prompt mix, user demographics, contamination, and model familiarity.
Source coverage
Publication cadence, public data availability, and source notes for each external source.
