AI · Article
Gemini 4 Carbon Tops MathArena, But Costs 15 Times More Than GPT-6.1 Sol
The Decoder ·
At a glance
Google's Gemini 4 Carbon has become the first non-OpenAI model in a while to top the MathArena math benchmark, pushing GPT-6.1 Sol out of first place. Claude Fable 5.2 took second place. OpenAI's newer models are not yet public.
Key points
- Gemini 4 Carbon took first place on the MathArena math test, the first non-OpenAI model in a while to do so.
- Claude Fable 5.2 placed second, and GPT-6.1 Sol dropped to third.
- Gemini 4 Carbon often runs out of time and rarely gives up, once running a script for more than eleven hours.
- With the test settings used, Carbon costs about 15 times as much as GPT-6.1 Sol, and Fable 5.2 about 40 times as much.
Why it matters
It shows that the most capable math AI can also be far more expensive to run, which matters for anyone choosing a model.
Original source
Gemini 4 Carbon takes the MathArena crown but costs 15 times more than GPT-6.1 Sol
The Decoder · Nov 15, 01:48 KST
Read on The Decoder ↗AI summarized the original article. Check the original for accuracy. Copyright belongs to The Decoder.
More from this field
- Nvidia's Security Team Under Scrutiny After CEO Jensen Huang's Exit
Nvidia's longtime security chief and product security head left in August, and CEO Jensen Huang left a few months later amid questions about his relationships. Reports say the security team's role in these events is unclear.
TechCrunch · Nov 15, 09:08 KST - Traditional Northern Ireland firms turn to AI to stay competitive
Kukoon Rugs, a family-run carpet maker in Northern Ireland, uses AI across nearly every department, from finance to product design. The firm says it lets them grow parts of the business they could not expand before.
BBC News · Nov 15, 09:00 KST - Google Search Ad Growth Slows as AI Search Faces Monetization Test
Google's search ad growth slowed sharply in the third quarter, while Meta's ad sales grew much faster. Investors are worried that AI search may be taking users away from regular search results, and that AI answers are not yet attractive to advertisers.
디지털투데이 · Nov 15, 08:32 KST - Character.AI Bankruptcy and OpenAI's Sora Shutdown Fuel Growing AI Graveyard
Many AI products and startups are shutting down, changing direction, or falling short of expectations, according to a TechCrunch report. Character.AI filed for bankruptcy in October after lawsuits and a Disney copyright dispute, and OpenAI ended its Sora video-sharing platform in March. The article lists several other AI products that closed or were absorbed into larger services.
디지털투데이 · Nov 15, 08:32 KST