Tech Desk: Google’s highly anticipated Gemini 3.5 Pro AI model has been significantly delayed, causing Alphabet’s stock to tumble by approximately 4% on July 16. Initially previewed at Google I/O 2026 with a promised June launch, the advanced AI model is now stuck in development, reportedly due to its coding performance falling short of internal expectations.

A Bloomberg report highlights that the Gemini 3.5 Pro struggled with tasks like writing, understanding, and debugging code. Despite Google’s efforts in late June to fine-tune the model with more programming-related data, the improvements weren’t sufficient. The tech giant is now reportedly restarting its entire training process, indicating a more profound issue than initially thought.

This setback comes at a critical time as Google faces intense competition in the rapidly evolving AI landscape. Rivals such as OpenAI have already released their next-generation models, like GPT 5.6 Sol, boasting superior token efficiency in coding tasks. Anthropic has also launched Claude Fable 5 and Mythos 5, further intensifying pressure on Google to deliver a competitive product.

While Google stated it’s “currently testing 3.5 Pro” and aims to “ship quickly,” no new timeline for the public release has been provided. The ongoing delay suggests users may have to wait considerably longer for the upgraded Gemini model, potentially impacting Google’s standing in the fiercely competitive AI race.