Alphabet Delays Gemini 3.5 Pro AI Model Over Coding Issues

What happened

Google announced Gemini 3.5 Flash at its I/O developer conference in mid-May and stated that the Pro version would arrive in June. That deadline passed without an update. According to reports, Google is taking additional time to improve Gemini 3.5 Pro's capabilities, particularly in coding. In late June, the company updated the model's training data in an attempt to enhance these skills, but the results were disappointing. Google confirmed it is currently testing 3.5 Pro, an upgraded Flash model, and other models with partners. The model remains in limited enterprise preview through Vertex AI, with a rumoured target date of mid-July for broader availability. Gemini 3.5 Pro is designed to feature a 2-million-token context window and a Deep Think reasoning mode for complex problem-solving tasks. Meanwhile, Gemini 3.5 Flash, the lighter sibling launched in May, is benchmarked as four times faster than the previous 3.1 Pro model.

Why it matters

The delay highlights potential challenges in Google's AI development process as it competes with rivals like Anthropic and OpenAI. Coding performance is a critical capability for enterprise AI applications, and Google's struggle to meet its own standards in this area could affect its competitive position. The setback has reportedly frustrated engineers, researchers, and managers inside the company, with some worried that competitors are pulling ahead. Additionally, reports indicate that 75% of all new code at Google is now AI-generated and approved by engineers, up from 50% last fall, underscoring the strategic importance of coding capabilities for the company's own operations. The departure of several senior researchers from Google DeepMind to Anthropic during this critical development phase adds further complexity to the execution timeline.

Bigger picture

The delay reflects broader challenges facing large technology companies as they race to develop and deploy advanced AI models. Google's sheer size may be working against it, with multiple teams across Google Cloud, DeepMind, Android, and other divisions building separate AI coding tools. Several layers of stakeholders are involved in preparing models for release, and employees reportedly face competition for computing power when trying to use AI internally. This contrasts with smaller, more focused competitors like Anthropic and OpenAI, which may be able to move more quickly. The enterprise AI market is particularly relevant for applications requiring sophisticated reasoning capabilities, including parsing complex codebases and analysing lengthy datasets. Google's ability to deliver on its promised improvements in coding performance will be closely watched by enterprise customers and developers.

What to watch

The key upcoming signal is whether Gemini 3.5 Pro launches in mid-July as rumoured and whether it delivers meaningfully better coding performance to justify the additional development time. Investors should monitor how the model's 2-million-token context window and Deep Think reasoning mode perform in real-world enterprise applications once broadly available. Any further delays or disappointing benchmark results could raise questions about Google's execution in AI development. The company's ability to retain senior AI researchers and resolve internal coordination challenges will also be important. Finally, watch for comparisons between Gemini 3.5 Pro and competing models from Anthropic and OpenAI in coding benchmarks and enterprise adoption rates.