Gemini 3.5 Pro Delayed by Months
新智元 · wechat · 2026-07-17
Bloomberg reports that Google's highly anticipated Gemini 3.5 Pro (codenamed Cappuccino) has been delayed by months, primarily due to internal dissatisfaction with its coding capabilities. Google reportedly scrambled to adjust training data for a final push last month, but results remained underwhelming.
The article notes that internal bureaucracy, resource tug-of-war across product lines, and engineer distrust of AI-generated code are all slowing down model progress. Simultaneously, internal AI coding tools are facing compute capacity shortages. The author frames this as an industry-wide issue, suggesting that the marginal returns of scaling compute and data are diminishing, and frontier model iteration is hitting a plateau.
Related event: Gemini 3.5 Pro Delayed by Months Over Coding Performance(12 posts)→
More from Companies & People
- Hacettepe University launches a new AI and Data Engineering department — aykuterdemml · 2026-07-23
- AWS launches Quick autonomous agents for CRM, email, Slack, and follow-ups — emmanuelvivier · 2026-07-23
- Sakana AI expands NVIDIA tie-up to put Nemotron models into its multi-agent Fugu system — hardmaru · 2026-07-23
- Journalists Urge OpenAI Staff to Whistleblow on Internal Misalignment Incidents — RebeccaBellan · 2026-07-23
- US Threatens Sanctions After White House Accuses Moonshot of Distilling Anthropic's Model — TechCrunch AI · 2026-07-23
- Tesla says Q2 revenue hit $28.2B, putting it on a $113B annualized run rate — ChrisGPT · 2026-07-23