Sina Weibo's Tiny 3B AI Model Claims to Rival Google, OpenAI, and Anthropic on Reasoning Benchmarks, Sparking Industry Debate

Sina Weibo's tiny 3-billion-parameter AI model, VibeThinker-3B, is shaking up the industry by claiming to rival Google, OpenAI, and Anthropic on key reasoning benchmarks, but skeptics warn the results may reflect 'benchmaxxing' rather than real-world capability.

• Sina Weibo's VibeThinker-3B, a 3-billion-parameter language model, claims to match or outperform AI giants like Google DeepMind, OpenAI, and Anthropic on key reasoning benchmarks, including a 94.3 score on AIME 2026, rivaling models hundreds of times its size.
• The AI community is sharply divided over the results, with many researchers accusing the team of 'benchmaxxing,' citing real-world tests where the model fails basic practical tasks, while others acknowledge the engineering achievement of compressing strong reasoning ability into such a small model.
• The findings challenge the AI industry's scaling hypothesis by suggesting that verifiable reasoning can be compressed into compact models, potentially enabling cheap, fast reasoning engines to run on consumer hardware and reshaping how the industry approaches model design and deployment.