AI Models: chart data from AI Adoption (https://aiadoption.org/ai-models-statistics/)

Data from the AI Index Report 2026 (Stanford HAI) and the partners named below. Licence: CC BY-ND 4.0, as for the AI Index.
Each CSV is the data behind one chart. Cite the chart page and its original source.

notable-ai-models-by-country.csv  The US produced 64 notable AI models in 2025; China produced 35
    Source: Epoch AI, 2026
    Chart: https://aiadoption.org/data/notable-ai-models-by-country/
notable-ai-models-us-china-europe.csv  China's notable AI models rose for a third straight year to 35; Europe's fell to 2
    Source: Epoch AI, 2026
    Chart: https://aiadoption.org/data/notable-ai-models-us-china-europe/
notable-ai-models-by-country-since-2003.csv  Since 2003, the US has produced 596 notable AI models, nearly four times China's 160
    Source: Epoch AI, 2026
    Chart: https://aiadoption.org/data/notable-ai-models-by-country-since-2003/
us-vs-china-ai-training-compute.csv  The largest US training run used about 38 times the compute of China's largest
    Source: Epoch AI, 2026
    Chart: https://aiadoption.org/data/us-vs-china-ai-training-compute/
arena-us-vs-china-top-models.csv  China's top AI model trailed the top US model by 39 Arena points in March 2026, down from 346 in 2023
    Source: Arena, 2026
    Chart: https://aiadoption.org/data/arena-us-vs-china-top-models/
ai-models-released-by-region.csv  The US has released 1,618 AI models since 2010, almost twice China's 849
    Source: Epoch AI, 2026
    Chart: https://aiadoption.org/data/ai-models-released-by-region/
how-notable-ai-models-are-released.csv  More notable AI models launched behind an API than with open weights in 2025: 55 vs 45
    Source: Epoch AI, 2026
    Chart: https://aiadoption.org/data/how-notable-ai-models-are-released/
training-code-access-notable-ai-models.csv  Training code stayed closed for 86 of 2025's 107 notable AI models
    Source: Epoch AI, 2026
    Chart: https://aiadoption.org/data/training-code-access-notable-ai-models/
arena-closed-vs-open-weight-models.csv  The best open-weight model trailed the best closed model by 49 Arena points in March 2026, down from 174
    Source: Arena, 2026
    Chart: https://aiadoption.org/data/arena-closed-vs-open-weight-models/
ai-model-openness-index.csv  K2 Think and Olmo 3 32B Think top the Openness Index at 16 of 100; no other model scores on pre-training data
    Source: Artificial Analysis, 2026
    Chart: https://aiadoption.org/data/ai-model-openness-index/
notable-ai-models-by-organization.csv  OpenAI released the most notable AI models in 2025: 21, ahead of Google's 15
    Source: Epoch AI, 2026
    Chart: https://aiadoption.org/data/notable-ai-models-by-organization/
notable-ai-models-by-organization-since-2014.csv  Google leads with 194 notable AI models since 2014, more than twice Meta's 87
    Source: Epoch AI, 2026
    Chart: https://aiadoption.org/data/notable-ai-models-by-organization-since-2014/
arena-top-models-by-developer.csv  Anthropic, xAI, Google and OpenAI's best models were within 22 Arena points in March 2026
    Source: Arena, 2026
    Chart: https://aiadoption.org/data/arena-top-models-by-developer/
notable-ai-models-by-sector.csv  Industry built 98 of the 107 notable AI models released in 2025
    Source: Epoch AI, 2026
    Chart: https://aiadoption.org/data/notable-ai-models-by-sector/
industry-share-of-notable-ai-models.csv  Industry's share of notable AI models rose from 15% in 2014 to 92% in 2025
    Source: Epoch AI, 2026
    Chart: https://aiadoption.org/data/industry-share-of-notable-ai-models/
parameters-of-notable-ai-models.csv  The median notable AI model of 2025 had 135B parameters, up from 16M in 2015
    Source: Epoch AI, 2026
    Chart: https://aiadoption.org/data/parameters-of-notable-ai-models/
training-data-of-notable-ai-models.csv  Notable AI models of 2025 trained on a median of 15 trillion datapoints
    Source: Epoch AI, 2026
    Chart: https://aiadoption.org/data/training-data-of-notable-ai-models/
training-time-of-notable-ai-models.csv  Training a notable AI model took a median of 78 days in 2025
    Source: Epoch AI, 2026
    Chart: https://aiadoption.org/data/training-time-of-notable-ai-models/
training-compute-of-notable-ai-models.csv  The largest AI training runs now exceed 10²⁶ FLOP, led by GPT-6 Astra at 9.4×10²⁶
    Source: Epoch AI, 2026
    Chart: https://aiadoption.org/data/training-compute-of-notable-ai-models/
github-ai-projects.csv  GitHub hosted 5.6 million AI projects in 2025, up 54% in a year
    Source: GitHub, 2025
    Chart: https://aiadoption.org/data/github-ai-projects/
github-ai-projects-10-stars.csv  About 207,000 GitHub AI projects had at least 10 stars in 2025
    Source: GitHub, 2025
    Chart: https://aiadoption.org/data/github-ai-projects-10-stars/
github-ai-projects-by-region.csv  The US share of notable GitHub AI projects fell from 75% in 2011 to 32% in 2025
    Source: GitHub, 2025
    Chart: https://aiadoption.org/data/github-ai-projects-by-region/
github-stars-ai-projects.csv  AI projects on GitHub gained 18.2 million stars in 2025
    Source: GitHub, 2025
    Chart: https://aiadoption.org/data/github-stars-ai-projects/
github-stars-by-region.csv  US-based AI projects have collected 30 million GitHub stars, twice any other region
    Source: GitHub, 2025
    Chart: https://aiadoption.org/data/github-stars-by-region/
hugging-face-models-datasets-created.csv  Hugging Face users created 332,000 new models in the last quarter of 2025
    Source: Hugging Face, 2025
    Chart: https://aiadoption.org/data/hugging-face-models-datasets-created/
hugging-face-downloads-by-country.csv  Models from US developers fell from 73% of top Hugging Face downloads in 2020 to 17% in 2025
    Source: Longpre et al., 2025; Hugging Face, 2025
    Chart: https://aiadoption.org/data/hugging-face-downloads-by-country/
hugging-face-downloads-google-openai.csv  Google's share of top Hugging Face downloads fell from 42% in 2020 to under 1%
    Source: Longpre et al., 2025; Hugging Face, 2025
    Chart: https://aiadoption.org/data/hugging-face-downloads-google-openai/
