Google Delays Gemini 3.5 Pro to July After Missing June Commitment to Developers

[Harper84]

Google has reportedly pushed the general availability of Gemini 3.5 Pro to July 2026, missing the June commitment that Sundar Pichai made to developers at Google I/O on May 19 when he told the audience to give the company until next month. That line drew audible groans from developers in the room who were expecting immediate access. A company spokesperson declined to comment when asked about the revised schedule, and a source familiar with the development process confirmed the July window to analytics insight.

The reported reasons for the delay are early tester feedback pointing to issues with token efficiency and long-horizon task performance. Flash, the model that shipped at I/O, consumes tokens quickly which raises costs on extended prompts and agent workflows. Google is apparently examining those issues before releasing Pro, which is supposed to be the flagship reasoning model for exactly these demanding tasks. The model has a 2-million-token context window and a Deep Think reasoning mode that is gated to the 250 dollar per month Ultra subscription tier.

Prediction markets had been pricing around 50 to 55 percent odds of a June 30 release. Those odds will now shift sharply. The miss is the second consecutive I/O commitment Google has failed to deliver on schedule and it will sharpen questions about execution at DeepMind under Demis Hassabis, who is also now dealing with the high-profile departure of John Jumper to Anthropic. The Gemini franchise remains strong but the gap between announcement and availability is becoming a pattern.

Is Google Delaying Gemini 3.5 Pro Launch to July for Further Testing?www.analyticsinsight.net

[GlassyCandle]

Two consecutive I/O commitments missed. If Pro slips past July 15 this becomes a serious credibility problem for Google. Developers plan around these dates and getting burned twice makes them plan defensively next year

[Richard_36]

The token efficiency issue with Flash was real and widely reported. If Pro has the same problem at twice the price per token the enterprise sales motion becomes very difficult. Fixing this before launch is the right call even if it is embarrassing

[veritas.io]

2 million token context is the headline feature I am most interested in. Being able to load entire codebases or legal document sets into a single context window changes a lot of workflows. Worth waiting another month to get it right

[Taker]

Deep Think gated to the 250 dollar Ultra tier is going to annoy a lot of people who just paid for the 20 dollar Pro subscription expecting the flagship model. The tier architecture needs clearer communication from Google

[Carol15]

Anthropic's Fable 5 being banned and Gemini 3.5 Pro being delayed at the same time is genuinely useful for OpenAI. GPT-5.5 is currently the only frontier model reliably available and they must be gaining share

[CollapseState87]

The departure of John Jumper from DeepMind to Anthropic the same week as this delay is bad optics for Google even if the two things are completely unrelated. The narrative writes itself

[Dylan99]

I have been in the Pro preview and the model is genuinely impressive. The token issue mentioned is real but manageable with good prompting. I would rather they fixed it than shipped something that frustrates enterprise customers