What changed

Google's Gemini 4 Carbon has become the first non-OpenAI model to top the MathArena math benchmark, displacing GPT-6.1 Sol from the top position. Claude Fable 5.2 secured the second spot. The model's performance in the benchmark is noted as a significant achievement, though it comes with notable cost implications.

What this means in practice

Gemini 4 Carbon's ability to solve complex mathematical problems has been highlighted, with the model demonstrating persistence in task completion, sometimes running scripts for over eleven hours without interruption. However, this performance is achieved at a cost that is approximately 15 times higher than GPT-6.1 Sol when using the reasoning settings from the test. The model's release date is November 10, 2026, with an expected cost of $15.70 per unit.

Limitations and availability

The model's most common failure mode is exceeding the time limit, which could be a concern for users requiring efficient processing. While the model is available for use, the cost and performance trade-offs should be considered. The benchmark results indicate that Gemini 4 Carbon is currently the top performer in the MathArena test, but the high cost may limit its accessibility for some users.

Source facts for Gemini 4 Carbon tops MathArena but costs 15 times more than GPT-6.1 Sol: OpenAI's models, which recently allegedly solved demanding equations like the Navier-Stokes equations, aren't publicly available yet. Gemini 4 Carbon's most common failure mode is exceeding the time limit. The model also almost never gives up on a task, a trait that can be read as either positive or negative.
Original AINewsTrend explainer based on the cited source. This graphic summarizes source statements; it is not an independent benchmark.

Sources and further reading

  1. thedecoder: Gemini 4 Carbon takes the MathArena crown but costs 15 times more than GPT-6.1 Sol