Google has delayed the launch of its artificial intelligence (AI) model, Gemini 3.5 Pro, from June to July.

Business Insider reported on Tuesday that Google adjusted the schedule to further reflect feedback from early test users and real-world use cases.

Google had signalled the launch of Gemini 3.5 Pro at its I/O developer event in May, but said at the time that the model was not yet ready. Sundar Pichai (순다르 피차이), Google's chief executive officer, referred to the launch timing as "next month" at the event. The planned June release has now slipped into July.

The adjustment largely reflects an effort to improve the model's completeness. An official said Google is spending time collecting more real-world use cases from early users. Gemini 3.5 Pro has been provided to some users through Google's Antigravity platform and the AI benchmarking site LMArena.

Google is also reviewing the scope of how the model can be used in this process. Gemini 3.5 Pro is expected to improve long-term multi-step task execution and agent operation performance. It is focused on handling multi-step work and automated AI use rather than simple response performance.

Google is also reflecting feedback on the recently introduced Flash 3.5 model in Gemini 3.5 Pro. The official said criticism of Flash 3.5 included that it consumes tokens too quickly. This is read as the background to Google tuning cost efficiency and real-world usability even if it delays the release of the higher-end model.

The delay itself is short, but the timing pressure is not small. Competition over model performance in the AI industry is intensifying quickly, and coding has become the first major enterprise use case for modern AI. Gemini 3 delivered better-than-expected results last year, but assessments say Anthropic and OpenAI continue to lead in coding.

In this situation, Gemini 3.5 Pro is likely to be Google's next key bet. Google is seeking to raise its capabilities for long-term task execution and agent use with the new model. As enterprise AI demand expands around coding and automation, the key question is whether the July release will lead to real performance improvements.