Google reclaims AI top spot with Gemini 4 Carbon, surpassing OpenAI and Anthropic

Google, once criticized for falling behind in the AI race, has reclaimed the top position with its new model Gemini 4 Carbon, surpassing the latest offerings from OpenAI and Anthropic in an AI performance benchmark. The company plans to use the milestone as a springboard to shift its core services — including search, smartphones and web browsers — toward an AI agent-centered model.

Google unveiled Gemini 4 Carbon on Wednesday (local time), according to Bloomberg and other outlets. The model is the successor to Gemini 4 Argon, which was introduced two weeks earlier.

AI evaluation firm Artificial Analysis announced that Gemini 4 Carbon scored 59 on its AI Intelligence Index (AAII), placing it first among all AI models — ahead of the latest releases from both Anthropic and OpenAI.

Google had faced criticism in recent months over delays in releasing its top-tier "Pro" model and lagging behind rivals in AI coding performance, at one point dropping out of the AAII top 10. The new release is being seen as a demonstration that the company has restored its technical edge.

According to Google, Gemini 4 Carbon showed particularly strong performance across cybersecurity. The model also recorded significant improvements in alignment — the ability of AI to act in accordance with human intent — and in safety.

Google CEO Sundar Pichai said at Alphabet's third-quarter earnings conference call Wednesday that the company would accelerate its shift to AI agents, using Gemini 4 Carbon as the foundation.

"When you see the new model's ability to automate and execute complex tasks at high quality, it's clear that the agentic capabilities everyone imagined are becoming real," Pichai said.

Pichai added that Google is distilling the Carbon model to efficiently deliver an agent experience to about 1 billion users.

Distillation is a technique in which the outputs of a large AI model are used as training data to develop a smaller, lighter model that achieves comparable performance. Because reducing power consumption and computing costs is critical for AI agents to execute complex tasks quickly, Google plans to use the high-performance Carbon model as a base to improve efficiency.

"Agents will serve as a unifying layer across our products," Pichai said. "We will expand the agent experience beyond search and the Gemini app to Chrome, Android, Workspace and more."

However, the transition to an AI-first business has also exposed a slowdown in Google's traditional search advertising revenue.

Alphabet's search advertising and related businesses grew about 12 percent year-on-year in the third quarter — down 5 percentage points from the 17 percent growth recorded in the previous quarter.

Alphabet Chief Business Officer Philipp Schindler attributed the slowdown to exchange rate effects as well as the integration of AI features into the search service.

Schindler said a growing share of Google search users are turning to "AI Overviews" and "AI Mode" instead of traditional search results, reducing the reach of conventional search ads.

Google said it plans to improve profitability by expanding ad placements within its AI search service.

"We are expanding ads in AI search results to more users and queries," Schindler said, adding that the monetization efficiency of AI-based advertising is expected to improve.

With Gemini 4 Carbon returning Google to the front of the AI performance race, the company's next challenge is translating that technical lead into tangible business results — particularly whether it can accelerate the rollout of AI agents while also reversing the slowdown in its core search advertising business.