Despite widespread predictions that AI development would hit a wall, Stanford's annual AI Index report confirms that top models continue to improve at an accelerating pace. People are adopting AI faster than they adopted the personal computer or the internet.
As of April 2026, Anthropic leads global AI model rankings, trailed closely by xAI, Google, and OpenAI. The benchmark SWE-bench Verified, measuring software engineering AI, saw top scores jump from around 60% in 2024 to nearly 100% in 2025.
On Humanity's Last Exam — the toughest multi-domain benchmark available — top models now score above 50%, compared to just 8.8% in early 2025. The improvement curve shows no flattening, defying sceptics who argued that scaling laws would hit diminishing returns by now.
The report also highlights a dramatic shift in AI research funding. Private investment in AI companies reached $338 billion globally in 2025, a 72% year-on-year increase. The United States captured 61% of that total, followed by China at 17% and the United Kingdom at 8%.
Perhaps most striking is the adoption data. Generative AI tools have been adopted by 53% of the global population within three years of launch — faster than any technology in recorded history. The estimated value of generative AI tools to US consumers alone reached $172 billion annually by early 2026.
Stanford's researchers emphasise that the speed of improvement is itself accelerating. Models released in Q1 2026 outperform models released in Q4 2025 by margins that would have taken a full year to achieve in 2023. This compression of progress is reshaping every forecast.
The implications for the global economy are profound. The report estimates that AI could add between $13 trillion and $22 trillion to global GDP by 2030, with the range widening significantly compared to last year's estimate due to the pace of capability gains.