Google Gemini 3.5 Pro delayed over coding performance issues

Key Takeaways

• Google's Gemini 3.5 Pro missed its June 2025 release target due to underperforming coding capabilities
• Internal teams updated training data in late June, but results still fell short of expectations
• Alphabet shares dropped nearly 3% following Bloomberg's report on the delay

Google's most powerful AI model, Gemini 3.5 Pro, is now months behind schedule. The model was supposed to ship in June 2025, but coding performance fell short of internal benchmarks, according to a Bloomberg report that sent Alphabet shares down nearly 3%.

CEO Sundar Pichai announced the June target at Google I/O in May. That deadline has come and gone. Bloomberg, citing ten current and former employees, reports that engineers, researchers, and managers inside Google are worried as OpenAI and Anthropic push ahead with models that outperform Gemini on key tasks.

What went wrong with Gemini 3.5 Pro?

The issue centers on coding capabilities. For enterprise customers and developers, code generation is one of the most valuable AI applications. It is also one of the hardest to get right. Late last month, Google updated the training data for Gemini 3.5 Pro in an attempt to improve these capabilities. The results fell short of expectations.

A Google spokesperson confirmed the model is still in testing. "We're currently testing 3.5 Pro, an upgraded Flash model, and other models with partners, and we're productively engaged with the U.S. government," the spokesperson told Reuters. "We're shipping quickly across a wide range of models while keeping them highly cost-effective for customers."

The company did not provide a new release date.

How does this affect the AI race?

Google finds itself in an uncomfortable position. OpenAI launched GPT-5.6, its most advanced model, last week. The release came after a brief delay prompted by U.S. government requests over national security concerns. Anthropic, meanwhile, faced its own regulatory friction. Its Mythos 5 and Fable 5 models were temporarily disabled for all users following a June 12 export control order. Anthropic added safeguards, and the curbs were lifted in late June.

Google has not faced similar regulatory blocks, but the technical delay puts distance between Gemini and its competitors at a moment when enterprise adoption is accelerating. Companies choosing an AI provider for code assistance, document processing, or customer-facing applications often evaluate benchmark performance. Falling behind on those benchmarks means falling behind in sales conversations.

Why coding performance matters most

Coding is not just a feature. It is a proxy for reasoning ability, instruction-following, and practical utility. Developers are among the earliest and most intensive users of AI assistants. They can immediately tell when a model hallucinates a function that does not exist or writes code that fails to compile.

For Google, which sells Gemini-powered tools through Google Cloud and integrates the model into Workspace products, coding quality directly affects revenue. Enterprise customers pay for reliability. A model that underperforms on code generation is a hard sell to engineering teams already comfortable with GitHub Copilot or Claude.

What happens next?

Google says it is testing Gemini 3.5 Pro with partners and the U.S. government. That suggests a launch is still planned, just not imminent. The company has other models in the pipeline, including an upgraded Flash variant optimized for speed and cost. Whether these can close the gap with OpenAI and Anthropic depends on what ships and when.

For now, the delay is a signal. Building frontier AI models is harder than the pace of announcements suggests. Every lab faces tradeoffs between capability, safety, and speed. Google chose to hold rather than ship something incomplete. That is defensible. But it leaves customers waiting and competitors running.

Frequently Asked Questions

When was Google Gemini 3.5 Pro supposed to launch?

CEO Sundar Pichai announced a June 2025 target at Google I/O in May. That deadline has been missed.

Why is Google Gemini 3.5 Pro delayed?

Bloomberg reports the model's coding capabilities fell short of internal goals. Google updated training data in late June, but performance still did not meet expectations.

How did the market react to the Gemini delay news?

Alphabet shares dropped nearly 3% following the Bloomberg report.

Is Google still planning to release Gemini 3.5 Pro?

Yes. A Google spokesperson confirmed the model is being tested with partners and the U.S. government, though no new release date has been announced.

How does Gemini compare to OpenAI and Anthropic models?

According to Bloomberg's sources, some Google employees are worried that OpenAI and Anthropic models are outperforming Gemini, particularly as those rivals release new systems.

Source: Tech-Economic Times / ET