AI adoption, spending, usage patterns, benchmark signals, model availability, and model mix.

Can I reuse the data?

Yes. The public JSON and Markdown files are designed for citation targets, lightweight analysis, and dashboards. External report figures should cite the original source.

Why do the numbers differ?

AI sources measure different samples and time periods. Keeping notes close to each chart prevents surveys, benchmarks, usage studies, and model data from being mixed incorrectly.

AI Model Reports

Public model data and sources

Public catalog coverage, source types, pricing units, categories, and reusable JSON and Markdown files.

Organizational AI adoption

Stanford HAI

88%

organizations

The 2026 AI Index reports broad organizational adoption of AI.

Industry frontier-model share

Stanford HAI

>90%

2025 notable frontier models

Stanford AI Index reports most notable frontier models in 2025 came from industry labs.

Enterprise GenAI spend

Menlo Ventures

$37B

2025 spend

Menlo estimates U.S. enterprise generative AI spend reached $37B in 2025.

Enterprise spend growth

Menlo Ventures

3.2x YoY

growth

Estimated enterprise GenAI spend grew from $11.5B in 2024 to $37B in 2025.

Model market data

Data landscapeSources

Adoption, spend, public datasets, and TokenLab's current model coverage.

Trend reportCharts

Compute, context length, cost movement, adoption, and model mix in one visual report.

Research librarySources

Selected model research with source links and practical buying notes.

Reusable public data

Use these files for citation targets, lightweight analysis, or downstream dashboards. They include public model data and summary fields.

Download JSON/model-data/catalog.json Download JSON/model-data/latest.json Download JSON/model-data/trends.json Download JSON/model-data/research.json Read Markdown/model-data/summary.md

Source coverage

Sources: Stanford AI Index 2026; Menlo Ventures 2025 Enterprise GenAI report; OpenAI enterprise usage data. Consumer ChatGPT figures from OpenAI Signals exclude enterprise and Codex usage and should not be combined with enterprise figures.