aii.news
Builders14 Nov

Google's Gemini 4 Carbon tops math benchmark but costs 15 times more than OpenAI's GPT-6.1 Sol

a calculator and a pen sitting on top of a piece of paper
Photo: Aaron Lefler / Unsplash

Gemini 4 Carbon and Anthropic's Fable 5.2 have pushed GPT-6.1 Sol off the top of the MathArena math benchmark, though both cost much more to run. In testing, Gemini ignored the time limit and almost never gave up on a task.

Why it matters

Developers choosing a model for hard reasoning tasks must now weigh small accuracy gains against much higher costs.

aii.news picks and summarizes stories from trusted news sources with the help of AI. Every story links to the original reporting. How we work.