Particle.news
Download on the App Store

OpenAI Sounds 'Code Red' as Google’s Gemini 3 Outperforms on Benchmarks

Altman’s directive follows reports of Gemini 3 topping benchmarks with longer user sessions.

Overview

  • Sam Altman issued an internal “code red,” directing teams to drop other work and improve ChatGPT’s core performance and reliability.
  • Recent coverage reports Gemini 3 beating ChatGPT on most standardized tests, with high-profile praise including Salesforce CEO Marc Benioff saying he switched to Gemini.
  • Usage data cited by outlets shows ChatGPT at roughly 800 million users and Gemini near 650 million, with Similarweb indicating users spend more time per visit on Gemini.
  • Analyst notes signal mounting financial strain for OpenAI, including HSBC’s estimate that it may need about $200 billion over four years and Deutsche Bank’s projection of roughly $140 billion in cumulative losses before profitability.
  • Media reports say OpenAI is accelerating a GPT-5.2 release targeting December 9 to close the gap, though the company has not confirmed the timing or details.