Current State of AGI
Understanding where we are in the journey towards Artificial General Intelligence
Latest Developments
As of March 2026, four frontier models launched within a 23-day window: Gemini 3.1 Pro, GPT-5.4, Grok 4.20, and Mistral Small 4. Claude Opus 4.6 and Sonnet 4.6 lead in agentic coding. AI agents autonomously complete tasks lasting over 4 hours. Nearly half of code on GitHub is now written using AI tools.
GPT-5.4: 1M Context, Native Computer Use, 75% OSWorld-V (March 2026)Claude Opus 4.6 & Sonnet 4.6: 1M Context, Agentic Coding (Feb 2026)Gemini 3.1 Pro Dominates 13 of 16 Major Benchmarks (Feb 2026)Gemini 3.1 Flash Live: Highest-Quality Voice AI Model (March 26, 2026)International AI Safety Report: Capabilities Outstrip Safety Measures
Key Challenges
The February 2026 International AI Safety Report (100+ experts led by Yoshua Bengio) warns that 'AI capabilities are advancing at a rate that outstrips current safety measures.' Metaculus forecasters predict median AGI arrival by February 2028. The gap between capability and safety remains the defining challenge.