Google has reportedly delayed the public launch of Gemini 3.5 Pro after the flagship AI model fell short of internal expectations, particularly in coding performance. While the company continues testing the model with selected partners, reports suggest engineers are rebuilding key parts of the system to improve reasoning and software development capabilities.

Google Gemini 3.5 Pro delay highlights AI race

Google's highly anticipated Gemini 3.5 Pro has yet to reach general availability despite being announced during Google I/O 2026 with a planned June release. While Gemini 3.5 Flash launched on schedule, the more capable Pro model remains available only to a limited group of enterprise partners and testers. According to reports citing people familiar with the project, the release has slipped several times as engineers work to improve core capabilities before a wider rollout.

Google has acknowledged that Gemini 3.5 Pro is still being tested with partners alongside an upgraded Flash model and other AI systems. The company has not announced a revised public release date but says it remains focused on delivering reliable and cost-effective models for customers.

Google Gemini 3.5 Pro coding issues become key challenge

Reports indicate that coding performance has become the biggest obstacle for Gemini 3.5 Pro. During internal testing, the model reportedly struggled with software generation, recursive tool calling and complex multi-step reasoning, areas that have become increasingly important as AI assistants evolve into autonomous coding agents. Google is said to have retrained the model using updated datasets in late June, but the improvements reportedly fell short of expectations.

Because AI-powered programming has become one of the industry's most valuable commercial applications, coding benchmarks now play a major role in determining how enterprise customers evaluate competing models. Google faces growing competition from OpenAI, Anthropic and several emerging AI companies that continue releasing increasingly capable coding assistants.

Gemini 3.5 Pro ground up rebuild reportedly underway

According to multiple reports, Google's engineering teams concluded that some of the model's limitations could not be solved through additional fine-tuning alone. Instead, the company reportedly began rebuilding important components of Gemini 3.5 Pro from the ground up while preparing an upgraded version of Gemini Flash as an interim release. Although Google has not confirmed details of an internal rebuild, it has confirmed ongoing testing with selected partners, including enterprise customers and government organizations.

The reported delays have raised concerns among some Google employees who worry that competitors are moving faster in the rapidly evolving AI market. Analysts note that frontier AI development has become increasingly difficult as companies attempt to improve reasoning, lower inference costs and satisfy enterprise reliability requirements simultaneously.

Coding remains the defining AI benchmark

Unlike earlier generations of AI systems, today's flagship models are increasingly judged by how effectively they write, debug and execute software. Professional developers, enterprise customers and benchmark platforms compare coding performance almost daily, making software engineering one of the industry's most competitive categories.

For Google, Gemini 3.5 Pro represents more than another model release. It is expected to power products across Google Cloud, Workspace and other enterprise services, making reliability essential before a full rollout. Delaying the launch may disappoint users waiting for the model, but releasing an underperforming flagship could create even greater reputational risks in an increasingly competitive AI landscape.