Alphabet Inc. (NASDAQ:GOOGL) has nearly caught Anthropic in a demanding new AI benchmark, with its latest Gemini model delivering almost the same performance at roughly one-third the price.
The new benchmark from SemiAnalysis puts Google’s Gemini 4 Carbon just behind Anthropic’s flagship Claude Opus 5.5, but ahead of OpenAI’s GPT-6 Astra.
The result marks a significant advance for Google, which is challenging its biggest AI rivals on both performance and price.
Google Closes the Gap
Carbon scored 10.9% on the GDPx benchmark, compared with 12% for Anthropic’s Opus 5.5. At Google’s introductory pricing, Carbon cost $16.29 per task, against $51.41 for Anthropic.
Google’s previous Gemini 4 Argon managed just 7.3%, underscoring how quickly the company has improved its models.
“Gemini 4 Carbon is a real improvement from Google and the only model in the same ballpark as Opus and Astra,” SemiAnalysis researchers wrote Monday.
Carbon led in software engineering and biology, although the researchers said its biology advantage likely benefited from less restrictive safety filters.
With existing AI benchmarks becoming increasingly easy for leading models, SemiAnalysis designed GDPx to test the limits of their capabilities.
Working with 14 specialist partners, it developed 206 demanding tasks across 11 fields, including semiconductor design, finance and scientific research.
Even the best model scored just 12%, suggesting the new benchmark has plenty of room to measure future improvements.
Polymarket Traders Favor Google
Despite Anthropic winning GDPx, traders on Polymarket favor Google to finish 2026 with the world’s top AI model.
Google has roughly 57% odds, compared with 34% for Anthropic and 5% for OpenAI, in a market with more than $3.8 million traded.
The market resolves using the Arena leaderboard rather than GDPx.
The Stakes for Google
Google expects to spend $195 billion to $205 billion on capital expenditures this year, putting pressure on the company to show investors a return on its enormous AI investment. Concerns about that spending contributed to its recent stock selloff.
Google Cloud revenue surged 103% to $30.7 billion in the third quarter, and Carbon offers another encouraging sign. By approaching Anthropic’s performance at a fraction of the price, Google could strengthen its appeal to paying AI customers.
Carbon is not yet widely available, and its introductory pricing may change. But its early results suggest Google’s heavy investment is helping it build models that can compete for a larger share of the AI market.
Image: Shutterstock
© 2026 Benzinga.com. Benzinga does not provide investment advice. All rights reserved.
