Google Delays Gemini 3.5 Pro Launch for a Second Time

Gemini 3.5 Pro, announced by Google at its I/O conference in May as an answer to Claude and GPT in the race for the best agentic models, still hasn't reached the broader public. The launch has slipped from June to July, reportedly because of problems flagged by the company's own testers.

On stage at I/O, Sundar Pichai said Gemini 3.5 Pro would run up to four times faster than the previous Gemini 3.1 Pro generation and would beat rivals on reasoning benchmarks. The announcement itself got a cool reception from the audience once it became clear the model wouldn't ship right away.

I know you can't wait to get your hands on it. Give us until next month - Sundar Pichai, CEO of Google

Why Pro stalled while Flash shipped

Google chose to release the lighter Gemini 3.5 Flash variant right away, leaving the Pro version in limited internal testing and preview access for select corporate clients on Vertex AI. It was those clients who reported that the model performs worse than expected on tasks requiring many reasoning steps and sustained autonomous work, exactly the area where Google wanted to build an edge over its rivals.

Token consumption in the already available Flash model turned out to be an additional headache. Some users reported the model burning through far more tokens than expected on elaborate prompts and long workflows, meaningfully raising the cost of using it in business settings. Google is working to fix that issue before deciding to roll it out at scale in the Pro version.

Competitive pressure mounts

The delay hits Google at a moment when competition for corporate customers is intensifying month by month. Sensor Tower data shows ChatGPT holding a clear lead in market share, while Anthropic's Claude Mythos 5 dominates coding benchmarks, scoring 80.3 percent on SWE-Bench Pro versus 55.1 percent for Gemini 3.5 Flash. For Google, which wants to sell Gemini 3.5 Pro as a tool for complex corporate analysis and multi-step agentic tasks, that performance gap is a serious image problem.

The company has not officially commented on the new release date, sticking to informal signals pointing to July 2026. The absence of an official confirmation makes it hard to tell whether the delay will be limited to a few weeks or drag on longer, as has happened before with some of Google's AI announcements.

What it means for business users

For companies planning Gemini-based deployments, including Google Cloud partners in Poland, the delay means continuing to rely on earlier model versions or temporarily turning to competing solutions from Anthropic or OpenAI for projects that require long agentic tasks. History shows that in such a fast-moving industry, a few weeks of delay can meaningfully sway the technology choices of corporate customers, who today are looking above all for stability and predictable costs, not just raw computing power.

Sources: Gemini 3.5 Pro Slips to July; Google Sheds $225B (tech-insider.org), Google delays Gemini 3.5 Pro, releases 3.5 Flash at I/O (letsdatascience.com), Gemini 3.5 Pro Cleared for July Launch as Fable 5 Nears Return, GPT-5.6 Stays Locked (techtimes.com)