Gemini 3.5 Pro Delayed Due to Coding Issues

Google’s next-generation flagship AI model, Gemini 3.5 Pro, is running months behind schedule due to persistent challenges with its coding capabilities, according to a new report from Bloomberg. The company had previously signaled a June release window, but that deadline has now passed without a public launch.

At its I/O developer conference in May, Google officially unveiled the Gemini 3.5 family. While the lighter Gemini 3.5 Flash shipped the same day, the company’s blog post noted that the more powerful 3.5 Pro was already in internal use and that Google looked forward to releasing it “next month.” That post was dated May 19, making the target month June. Yet as of this writing, the Gemini API release notes contain no entry for a Gemini 3.5 Pro model.

Bloomberg reporters Julia Love and Davey Alba, citing sources familiar with the matter, revealed that the delay stems from underwhelming results after Google updated the training data for Gemini late last month in an effort to sharpen its coding performance. Ten current and former employees described growing frustration inside the company, with some worried that Google is losing ground as rivals Anthropic and OpenAI continue shipping models that outperform Gemini. These sources spoke on condition of anonymity due to the sensitivity of internal concerns.

A Google spokesperson told Bloomberg that the company is currently testing 3.5 Pro with partners, alongside an upgraded version of Flash. Still, the company has not announced a new public release date.

This isn’t the first time Google has acknowledged a gap in agentic coding , the ability for AI to autonomously generate and execute code. In May, SEJ reported that CEO Sundar Pichai admitted Google was “a bit behind” the frontier in this area, attributing the lag to the absence of a developer-facing coding product that would generate the training data competitors benefit from. SEJ’s June coverage of two senior departures from Google’s AI organization also cited Bloomberg reporting on internal concerns at DeepMind about what Google offers businesses building AI coding tools. Pichai’s own comments had already signaled that Google recognized its trailing position in agentic coding. Bloomberg’s latest report now indicates that broader coding performance is one of the key factors holding up 3.5 Pro.

The implications for Search are not immediate. At I/O, Google made 3.5 Flash the default model in AI Mode globally, and the company’s materials never suggested that 3.5 Pro would replace it. So a delayed flagship doesn’t directly affect the answers Search is generating today. What it does affect is Google’s model release timeline. The company missed the public rollout expectation it set in May and, as of now, has not announced a replacement date.

Looking ahead, Google’s statement to Bloomberg described testing with partners , which is not a rollout , and the company has not named a new month for release. Until it does, 3.5 Flash remains the only model Google has released in the 3.5 series. Any information circulating about when Pro might arrive traces back to reporting rather than to official word from Google.