Inception Labs' Mercury 2 Hits 90% on AIME 2026, Outperforms Google's DiffusionGemma

On Thursday (June 18), Inception Labs unveiled Mercury 2, claiming it as the world's fastest reasoning language model, generating approximately 1,000 tokens per second. According to the company's announcement, Mercury 2 scored 90% on AIME 2026 (American Invitational Mathematics Examination problems), compared to Google's DiffusionGemma at 69.1% on the same benchmark. The model also achieved 77% on GPQA, a PhD-level science benchmark. Augment Code, an AI coding-agent company, reported an 82% reduction in latency and 90% cost cut after swapping Mercury 2 for Anthropic's Claude Opus 4.7, maintaining output quality.
Disclaimer: The information on this page may come from third-party sources and is for reference only. It does not represent the views or opinions of Gate and does not constitute any financial, investment, or legal advice. Virtual asset trading involves high risk. Please do not rely solely on the information on this page when making decisions. For details, see the Disclaimer.
Comment
0/400
No comments