Google's AI models vanish from top 15 leaderboard, raising strategy questions
Google hasn't shipped a competitive AI model in months, falling behind rivals Sol and Fable.
Deep Dive
Google hasn't shipped a model recently that can compete with Sol or Fable. Its previous models were disappointing and unreliable. This has led to speculation about two possible strategies: either going all-in on on-device inference for its own products (a battle Apple might win due to better hardware and third-party model licensing), or being stuck in internal politics with no one shipping anything. What's actually going on?
Key Points
- Google's models have dropped completely out of the top 15 on recent AI leaderboards, missing out on the latest wave of competitive releases.
- Previous flagship models (e.g., Gemini) were deemed unreliable and disappointing, failing to keep pace with rivals.
- Two leading theories: a pivot to on-device AI (vulnerable to Apple's hardware advantage) or internal political deadlock stalling innovation.
Why It Matters
Google's retreat from frontier AI threatens its leadership in search, cloud, and consumer products—opening doors for competitors.