Google delays Gemini Pro as coding falls short of internal goals

Google is months behind schedule on its flagship Gemini Pro AI model, as the technology struggles to meet internal goals for coding performance, according to a Bloomberg report Source: thenextweb. The delay has caused Alphabet shares to drop 4% Source: CNBC.

What happened

Gemini Pro was expected to launch in June following its announcement at Google I/O in May Source: CNA. However, internal testing revealed that the model's coding capabilities lag behind rivals like OpenAI and Anthropic. Google updated training data in late June, but results were disappointing Source: thenextweb.

The reaction

Alphabet shares fell 3-4% on the news Source: CNA, CNBC. Engineers are leaving for AI labs like Anthropic, frustrated by Google's competitive position and internal competition for computing power Source: thenextweb. Some customers report mixed experiences: Figma praised the Flash model for speed and quality, while Platzi CEO Freddy Vega called it more expensive and slower than competitors, leading his team to switch to Anthropic Source: thenextweb.

Internal challenges

Google's AI efforts are fragmented across DeepMind, Cloud, and Android teams. Co-founder Sergey Brin has pushed for faster progress but faces resistance from those who believe human-written code remains important Source: thenextweb. Chief AI Architect Koray Kavukcuoglu is working to unify tools, and a new DeepMind team led by Sebastian Borgeaud focuses on coding. Google claims 75% of code is now AI-generated Source: thenextweb.

What's next

A Google spokesperson said the company is "shipping quickly across a wide range of models" and testing Pro, an upgraded Flash model, and others with partners Source: CNA, CNBC. With OpenAI and Meta recently releasing new models, the pressure on Google to close the gap intensifies.