TH
← Back
news 2026-04-23 · qbitai

GPT Image 2 Team Revealed: 13 Engineers, 4 Months, Half Chinese — The Tiny Team Behind AI's Biggest Image Breakthrough

GPT Image 2 Team Revealed: 13 Engineers, 4 Months, Half Chinese — The Tiny Team Behind AI's Biggest Image Breakthrough

OpenAI's GPT Image 2 — the model behind the viral Ghibli-style images and stunning infographics flooding the internet — was built by just 13 people in four months. And over half the team is Chinese.

The team's composition was recently revealed by Chinese tech outlet QbitAI, sparking intense discussion about the global AI talent war. Leading the effort is Cai Jun, a brilliant researcher from Wuxi, Jiangsu Province, who completed his PhD in the United States before joining OpenAI.

Under Cai's leadership, this skeleton crew accomplished what many companies with hundreds of engineers have failed to achieve — a generative image model so capable it crashed servers from sheer demand when released.

The revelation carries several implications:

**Small teams win big.** GPT Image 2 is arguably the most impactful consumer AI feature of 2026, yet it was built by a team smaller than most startup founding crews.

**The talent pipeline matters more than money.** China produces a disproportionate share of the world's top AI researchers, and American companies are the primary beneficiaries. This dynamic is now fueling heated debate in China about brain drain.

**Speed over scale.** Four months from concept to world-changing product challenges the assumption that breakthrough AI requires massive teams and years of development.

The story has reignited calls within China to improve domestic conditions for AI talent retention, even as U.S. immigration policies face scrutiny for potentially pushing talent away.

In an era where 13 people can reshape how billions create images, the real competitive advantage isn't compute or capital — it's people.

📄 Source

qbitai
Share: Facebook 𝕏
← Previous
Google Launches One-Stop AI Agent Platform for Ent
Next →
China's First Pure Inference GPU Unicorn Is Here —