Leak: Google to Launch Gemini 3.5 Pro on July 17
Leaked plans reported by tech outlets suggest Google DeepMind is set to unveil Gemini 3.5 Pro on July 17, after roughly six weeks of delay.
- Leaked plans suggest Google DeepMind will unveil Gemini 3.5 Pro on July 17.
- Unofficial reports mention a context window of up to 2 million tokens and a deeper reasoning layer.
- The launch is running roughly six weeks late, reportedly over token efficiency and coding issues.
- The date, pricing and technical figures stay unofficial until Google publishes documentation.
Leaked internal plans reported by technology outlets indicate that Google DeepMind is set to present a new language model called Gemini 3.5 Pro on July 17. It would succeed the current Pro line, and the reported date, if confirmed, comes after roughly six weeks of delay against the originally planned launch window.
According to the same unofficial sources, the model should bring a context window of up to two million tokens, a deeper reasoning layer aimed at more complex tasks, and noticeably stronger image and vector SVG generation. None of those details have yet been confirmed by official company documentation, so they should be read as a secondhand preview rather than a spec sheet.
What is known, and what stays speculation?
What can be said with confidence is that the leaked material exists and that several technology outlets are reporting the same date. Everything beyond that, the concrete performance figures, the cost of access and the exact list of new features, belongs to the realm of expectation until Google issues an official statement.
The delay is, according to those sources, the result of technical hurdles. Testers reportedly flagged problems with token efficiency and with code generation, which prompted the team to rework the model’s architecture more than once. That sequence is not unusual in large language model development, where the launch date slips until internal checks show stable results. It is worth keeping in mind that a delayed launch says nothing about a model’s final quality, only that the development cycle was not yet closed.
A context window of two million tokens, if confirmed, would mean the model can process very large amounts of text at once, such as extensive technical documentation or longer code bases. That is the direction the whole industry has been moving in for some time, so the announcement in itself is not a breakthrough, but a step in the expected direction.
Why now?
If the date holds, Gemini 3.5 Pro arrives during one of the busiest weeks in AI development so far. Rivals have already made their moves: OpenAI presented GPT-5.6 on July 9, and Grok 4.5 was released the same day. Three major names within just a few days shows how compressed the release cadence has become.
For users and development teams, a schedule like this has a practical side. When three models arrive almost simultaneously, comparisons happen in real time and strengths and weaknesses surface quickly. That is exactly why it pays to treat marketing claims with caution until independent tests and official measurements become available.
Until the official unveiling, all of the above remains at the level of leaked plans. The July 17 date, like the technical figures, will only be confirmed once Google publishes documentation. Until then, it is reasonable to follow the announcement with a healthy dose of reserve. What do you expect from the next generation of Gemini models, and does it really matter to you which company ships a new model first?
