Executive Summary
- 1OpenAI's o3 scored 87.5% on ARC-AGI — previous best was 55%
- 2Google shipped Gemini 2.0 with native tool use and agents
- 3Cursor hit $100M ARR in under 2 years, proving AI-native wins
12
Models Released
$4.2B
Funding ($M)
89
Tools Launched
3
Acquisitions
Winners
- OpenAI:o3 leapfrogged everyone on reasoning
- Cursor:$100M ARR, fastest dev tool growth ever
- Anthropic:Claude leads coding benchmarks
Losers
- Standalone AI wrappers:Models now do tool use natively
- Manual prompt engineering:Reasoning models self-correct
What Changed
Before
AI = pattern matching on training data
After
AI = genuine novel reasoning (maybe)
Implication: The 'just autocomplete' argument is dying
Watch Next Month
- OpenAI o3 public release and pricing
- Anthropic's response to reasoning models
- Whether Gemini 2.0 agents work in production
This Week in AI
5 storiesOpenAI's new reasoning model achieved near human-level performance on abstract reasoning tasks specifically designed to be hard for AI.
Gemini 2.0 can browse the web, execute code, and use external tools without separate API calls. It's multimodal-native with real-time video understanding.
The AI-powered code editor reached $100M annual recurring revenue, making it one of the fastest-growing developer tools ever.
Claude can now see your screen and control your computer to complete tasks. Currently in beta for developers.
The first provisions of the EU AI Act are now in effect, with high-risk AI systems facing new compliance requirements.