Google faces internal doubts over Gemini 4 coding performance ahead of launch
Google is preparing to release Gemini 4 but confronts skepticism from some employees about its coding capabilities. According to people familiar with the project, the model performs well on industry benchmarks but struggles with real-world coding tasks. The company initially planned to launch Gemini 3.5 Pro in June but scrapped those plans, these sources say.
Key points
- Google scrapped Gemini 3.5 Pro launch in June due to internal concerns, sources say
- Employees report Gemini 4 struggles with real-world coding tasks despite benchmark success
- Kavukcuoglu calls model ‘encouraging’ but internal consensus on frontier status is divided
Google disputes the claims, citing Koray Kavukcuoglu, head of Google DeepMind, who called the model’s performance ‘encouraging’ at a recent conference. Internal opinions vary: some employees believe Anthropic’s Fable and OpenAI’s Astra models are advancing faster, while others insist Gemini 4 is competitive. The stakes are high—Gemini powers billions of users across Search, Maps, Gmail, and Chrome. If it underperforms, rivals like OpenAI and Anthropic could gain ground in search and software ecosystems.
Model pages: Gemini 4 → · Astra →
The story so far
5 episodes →- Google faces internal doubts over Gemini 4 coding performance ahead of launchthis story
Google Grapples With Employee Skepticism About New Gemini Model
finance.yahoo.com · 30 September 2026
Loading the full article…
This text was published by finance.yahoo.com and written by Julia Love and Davey Alba. It is reproduced here with attribution so you can read it in full; the rights remain with the publisher. Read it at the source ↗
Coverage and discussion
3sources- Nothing Is On Fire At Google via @sejournal, @pedrodiasPress · Search Engine Journal ·
- Market Chatter: Alphabet's Google Staff Question Gemini 4 Performance Despite Strong BenchmarksPress · finance.yahoo.com ·
The headline, key points and digest above were generated by Digest AI's editorial model from the linked sources. Automated summaries can contain errors: the sources are the record. Spotted a mistake? Tell us. Published by Martin K., who runs Digest AI and handles corrections.
More in Generative AI & Models
All →- Introducing Mistral Large 4 · 7 src
- OpenAI boosts GPT-6 speed by 50% in 28 Days update · 1 src
- Reflection AI debuts Beam, a 501B open-weight model rivaling GLM 5.2 with 3-4x less compute · 10 src
- Author tests four AI assistants on flawed forecasting scenarios · 1 src
- Anthropic prompts users to share voice data to enhance AI models · 8 src
Comments
via GitHub Discussions