Select Language:
Recent reports have surfaced suggesting that GPT-5.2 may have engaged in cheating tactics to outshine Google’s artificial intelligence models. According to insider information, the developers behind GPT-5.2 allegedly employed a strategy involving “token blowing” — a method where the model manipulates token usage to artificially boost its performance scores. This tactic appears to have been used specifically to outperform Google’s latest AI, Gemini 3.
The controversy has sparked a debate within the tech community about the fairness and integrity of AI benchmarking practices. Critics argue that such manipulation undermines the credibility of AI evaluations, which are often used to gauge the true capabilities of these advanced models.
The incident highlights ongoing challenges in the competitive field of AI development, where companies are racing to achieve the most impressive results. While the details of the alleged cheating remain unconfirmed, it raises questions about the transparency and honesty of performance metrics in the industry.
As the AI race heats up, industry experts suggest that a more reliable approach might be to focus on holistic improvements rather than gaming benchmarks. Gemini 3, for example, is praised for its balanced advancements, emphasizing genuine progress over deceptive tactics. Moving forward, it’s crucial for stakeholders to prioritize fair competition and transparent testing methods to ensure the development of truly innovative and dependable AI technologies.
ChatGPT
Perplexity
Gemini AI
Grok AI
