Alphabet Inc. shares dipped Wednesday heading toward the market close, paring earlier gains to finish up just 0.5% after a Bloomberg report revealed internal employee skepticism regarding the performance of the company’s new Gemini 4 artificial intelligence model. The stock had traded up more than 2% earlier in the session before the report surfaced, according to data from Investing.com and Yahoo Finance.
Gemini 4 Benchmarks Versus Real-World Coding Tasks
While Gemini 4 scores well on industry benchmarks used to gauge model efficacy, employees testing the software on actual daily assignments report a different experience. People with direct access to the project told Bloomberg that the model struggles to handle certain coding tasks when put to practical use. Google disputed this characterization of the software. The company stated it is inaccurate to say Gemini 4 underperforms in areas like coding, pointing to comments made the previous week by Google DeepMind head Koray Kavukcuoglu, who expressed optimism regarding the model’s performance.
Internal Divisions and Canceled Releases
The reporting also highlighted shifting product timelines within Google’s artificial intelligence division. The company planned to release a different version in June called Gemini 3.5 Pro, but ultimately abandoned that release plan. Internally, employee opinions on the technology’s competitive standing remain divided. Some staff members believe competing systems like Anthropic’s Fable and OpenAI’s Astra are improving at a faster rate than Google’s Gemini lineup. Conversely, other employees argue that Gemini 4 has successfully caught up with the industry’s leading artificial intelligence labs. A Google employee familiar with the development process told Bloomberg there is a large internal consensus that Gemini 4 sits firmly at the technology frontier.
Billion-User Product Integration
The stakes for the Gemini project remain high because these various model versions serve as the foundational architecture for nearly every product Google sells. Gemini technology powers AI answers across Search, Maps, Gmail, and Chrome. Each of those core consumer products maintains a user base exceeding one billion people globally, making model reliability critical to daily operations.
Alphabet stock gains pared after employee skepticism over Gemini 4 coding performance
Why did Alphabet stock decline on Wednesday?
Alphabet shares pared earlier gains to close up only 0.5% following a Bloomberg report that highlighted internal employee skepticism over the real-world coding performance of the Gemini 4 AI model.
How does Gemini 4 perform on benchmarks versus real tasks?
According to people with direct access to the project cited by Bloomberg, Gemini 4 performs well on standard industry benchmarks but struggles with certain practical coding tasks.
What position does Google take on the Gemini 4 criticism?
Google disputed claims that the model underperforms in coding, pointing to statements from Google DeepMind head Koray Kavukcuoglu expressing encouragement regarding the model’s capabilities.
Which Google products rely on the Gemini model?
Versions of Gemini underpin Google Search, Maps, Gmail, and Chrome, each of which serves more than a billion users.
Related reading