Alphabet falls after reported delay to Gemini 3.5 Pro

Alphabet shares fell Thursday after Bloomberg reported that Google has pushed back the release of Gemini 3.5 Pro, its next flagship artificial intelligence model. CNBC market data showed Alphabet Class A stock down 4.61% at $353.84 at 3:25 p.m. EDT, reflecting investor concern over the pace of Google’s AI product rollout.

Bloomberg, citing people familiar with the matter, reported that Gemini 3.5 Pro is months behind schedule as Google works to lift the model’s performance. The report said the system’s ability to generate software code has fallen short of internal targets.

The delay places pressure on one of Alphabet’s most scrutinized growth initiatives. Advanced AI models have become a central competitive front for large technology groups, with coding tools emerging as a prominent commercial use case for developers and enterprise customers.

Google says testing is under way

Google introduced Gemini 3.5 Pro in May at its annual Google I/O developer conference. At the time, the company said the model was already in internal use and that a wider release would not come until the following month.

An Alphabet spokesperson told CNBC by email that the company is “shipping quickly across a wide range of models while keeping them highly cost-effective for customers.” The spokesperson added: “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.”

AI labs often release models in stages, first testing them internally and with selected partners before making them broadly available. That process allows companies to assess performance, cost and reliability before putting a system into wider commercial use.

Coding performance becomes a key battleground

Bloomberg reported that OpenAI and Meta have recently launched AI models that exceed Google’s current offerings in software code generation. Code generation has become one of the largest use cases for AI model providers including Anthropic and OpenAI, as well as Chinese AI labs such as Z.ai, which offer open-weight variants that developers can access without charge through the open-source ecosystem, according to CNBC.

Meta last week introduced Muse Spark 1.1. Alexandr Wang, Meta’s AI chief, described it as the company’s strongest model so far for agentic and coding work, CNBC reported.

OpenAI also released GPT-5.6 Sol last week. Chief Executive Sam Altman said that model is 54% more token efficient on agentic coding tasks, according to CNBC. Token efficiency is a metric AI companies use when describing the relationship between model performance and the cost of using a model at scale.

The reaction in Alphabet’s shares shows how closely public-market investors are watching the timing and competitiveness of major AI releases. Alphabet remains one of the largest companies investing in AI infrastructure and products, but Thursday’s move indicated that delays in flagship systems can weigh on sentiment as rivals promote new models aimed at developers and corporate users.

This story draws on original reporting from CNBC.