The 2025 AI Index Report highlights

Performance & Accessibility

• AI models have significantly improved on challenging benchmarks like MMMU, GPQA and SWE-bench.
• Smaller models are achieving high performance; for instance, Microsoft’s Phi-3-mini (3.8B parameters) matched larger models on MMLU.
• Inference costs have plummeted—from $20 to $0.07 per million tokens—making AI more accessible.

Global AI Landscape

• The U.S. leads in producing notable AI models (40 in 2024), but China is closing the gap with 15 models and leads in AI publications and patents.

Investment & Adoption

• U.S. private AI investment reached $109.1B in 2024, nearly 12 times China’s $9.3B.
• Generative AI attracted $33.9B globally, an 18.7% increase from 2023.
• AI adoption in businesses surged, with 78% of organizations reporting usage in 2024, up from 55% in 2023.

Regulation & Ethics

• AI-related incidents rose to 233 in 2024 a 56.4% increase from 2023 highlighting the need for robust AI governance.
• U.S. states are leading in AI legislation with 131 state-level laws passed in 2024 up from 49 in 2023.

Research & Development

• Training compute for notable AI models is doubling approximately every five months.
• AI’s share of computer science publications increased from 21.6% in 2013 to 41.8% in 2023.

Public Perception

• AI optimism varies globally: 83% in China view AI positively compared to 39% in the U.S.

Link of the full report: 2025 AI Index Report