Alphabet shares drop after reported delay in its Gemini 3.5 Pro AI model

Alphabet shares fell 4% on Thursday after Bloomberg reported, citing people familiar with the matter, that the launch of the company's flagship artificial intelligence model, Gemini 3.5 Pro, is running months behind the originally planned schedule. According to the report, the delay stems from the company's efforts to improve the model's performance ahead of its wider rollout.

The most notable point is that Gemini 3.5 Pro's code-generation capabilities reportedly fell short of the company's internal expectations. This comes at a time when rivals such as OpenAI and Meta have recently unveiled new models that, according to the report, outperform Google's current offerings specifically in software code generation.

Alphabet had announced Gemini 3.5 Pro in May, during its annual developer conference Google I/O, noting at the time that the model was already being used internally, though it would not be ready for wider deployment until the following month.

An Alphabet spokesperson responded to CNBC with an emailed statement saying the company is "rapidly shipping a wide range of models, while keeping them highly cost-effective for customers." The spokesperson added that they are currently testing 3.5 Pro, an improved Flash model and other models with partners, and that they maintain a productive working relationship with the U.S. government.

The article frames this episode within the intense competition among AI labs around code generation, one of the most significant use cases for model providers such as Anthropic and OpenAI, as well as for Chinese labs like Z.ai, which offer open-weight variants freely accessible to developers through the open-source ecosystem.

As examples of that competition, the article mentions that Meta unveiled last week its Muse Spark 1.1 model, described by its head of AI, Alexandr Wang, as the company's "strongest model to date" for agentic and coding work. Meanwhile, OpenAI released its GPT-5.6 Sol model last week, which its CEO, Sam Altman, said is 54% more token-efficient in agentic coding tasks, underscoring how AI labs are marketing their respective coding models as cost-effective relative to their performance.

The article also includes the assessment of John Belton, of Gabelli Funds, who in a televised interview argued that Google remains a "well-positioned" company, despite the stock drop triggered by news of the delay.