DeepMind admits Gemini lags frontier, promises future lead
Is this a scandal?
No longer — the story has resolved. Noise 29/100, cooling down, across 1 source.
Google will likely accelerate Flash model releases to maintain market presence while Gemini 3.5 Pro faces further delays because the team is reallocating compute resources toward the more ambitious Gemini 4 architecture.
Noise 29/100 — louder than 94% of tracked AI controversies.
Why it matters
Google's explicit admission of falling behind validates market concerns about its AI competitiveness and signals a strategic pivot toward Flash models while delaying flagship Pro releases.
Key points
- DeepMind SVP Koray Kavukcuoglu explicitly admitted current Gemini models fall below the industry frontier.
- Google provided no release timeline for Gemini 3.5 Pro, citing a strategic pivot to faster Flash model iteration.
- Kavukcuoglu confirmed Gemini 4 development is underway and described internal progress as promising.
- The executive framed missing agentic coding capabilities as a necessary learning process rather than a deficit.
- DeepMind leadership asserted that achieving frontier status remains their exclusive organizational priority.
- Kavukcuoglu characterized competitive leadership shifts between AI labs as natural industry ebbs and flows.
The story
DeepMind SVP Koray Kavukcuoglu acknowledged that Google’s current AI models perform below the industry frontier during a recent public appearance. Kavukcuoglu stated unequivocally that reaching the frontier remains the organization's sole priority despite current quality gaps. He confirmed that Gemini 3.5 Pro remains in development without a release timeline as teams prioritize faster-iterating Flash models. The executive also revealed parallel work on Gemini 4, describing it as an ambitious effort progressing well internally. Kavukcuoglu attributed the lack of agentic coding features to necessary learning phases rather than technical failure. He emphasized that competitive leadership naturally fluctuates between labs over time. This candid assessment addresses growing user criticism regarding Google's perceived stagnation in the generative AI race. The comments signal continued investment in frontier research despite near-term product delays.
Who's involved
Users and observers question Google's commitment to the frontier due to missing Gemini 3.5 Pro updates and agentic features.
DeepMind SVP acknowledges current model deficits but insists frontier leadership is inevitable through parallel research tracks.
Noise Level
The timeline
DeepMind SVP addresses Gemini delays publicly
Koray Kavukcuoglu admitted current models lag behind competitors and confirmed no imminent release for Gemini 3.5 Pro during a public discussion.
The full record
Sources & methodology
- twitter.com — twitter.com
Every claim above traces to these primary items. How we score →
The forecast
Google will likely accelerate Flash model releases to maintain market presence while Gemini 3.5 Pro faces further delays because the team is reallocating compute resources toward the more ambitious Gemini 4 architecture.
Forecast, not fact — an editorial estimate we score when this resolves.
That's the complete picture as of — nothing more to know right now. We'll update this page the moment it changes.
Join the Discussion
Discuss this story
Community comments coming in a future update
Be the first to share your perspective. Subscribe to comment.