%20--%3e%3cdefs%3e%3cstyle%3e%20.st0%20{%20fill:%20%23061b40;%20}%20.st1%20{%20fill:%20%23306af1;%20}%20.st2%20{%20fill:%20%235ce5cf;%20}%20%3c/style%3e%3c/defs%3e%3cg%3e%3cpath%20class='st0'%20d='M55,10.5h9v3h-9v9h12V7.5h-12v3ZM64,19.5h-6v-3h6v3Z'/%3e%3cpolygon%20class='st0'%20points='69%2016.5%2078%2016.5%2078%2019.5%2069%2019.5%2069%2022.5%2081%2022.5%2081%2013.5%2072%2013.5%2072%2010.5%2081%2010.5%2081%207.5%2069%207.5%2069%2016.5'/%3e%3cpolygon%20class='st0'%20points='95%2010.5%2095%207.5%2083%207.5%2083%2022.5%2095%2022.5%2095%2019.5%2086%2019.5%2086%2016.5%2095%2016.5%2095%2013.5%2086%2013.5%2086%2010.5%2095%2010.5'/%3e%3cpath%20class='st0'%20d='M40,1.5v21h11.6l1.4-1.4v-7.6h0c0,0-1.4-1.5-1.4-1.5l1.4-1.4V2.9l-1.4-1.4h-11.6ZM50,19.5h-7v-6h7v6ZM50,10.5h-7v-6h7v6Z'/%3e%3c/g%3e%3cpath%20class='st1'%20d='M23.1,24L14.7,4.8l-4.9,11.2h3.8l-1.8,4H3.3L12.1,0H2C.9,0,0,.9,0,2v20c0,1.1.9,2,2,2h21.1Z'/%3e%3cpath%20class='st2'%20d='M34,0h-16.8l10.6,24h6.2c1.1,0,2-.9,2-2V2C36,.9,35.1,0,34,0ZM32.5,20h-4V4h4v16Z'/%3e%3c/svg%3e)
At this year's I/O conference in May, Google made a big announcement that its flagship large model Gemini 3.5 Pro would be released in June, but it failed to meet the deadline. During the delay, several core AI researchers left the team to join competitors, and team morale dropped to an all-time low. Now, the latest news has finally given a clear timeline — Gemini 3.5 Pro is set to officially launch on July 17.
Interestingly, mid-July also happens to be the expected release window for the official version of the domestic large model DeepSeek V4. The face-off between these two flagship models from China and abroad will be one of the most anticipated highlights in the AI world this year.
Rejecting "toothpaste squeezing", a new foundation trained from scratch
Regarding the reason for the delay, insiders revealed that the Google team was not slacking off but instead used the extra time for a completely new pre-training, rather than fine-tuning based on the previous Gemini 2.5 Pro. This means that Gemini 3.5 Pro will be a truly significant upgrade, not an incremental improvement. Google chose to take the risk of delays to ensure the foundation was well-prepared, showing high expectations for this model.
According to leaks, the biggest breakthrough of the new model lies in the front-end generation field. The design aesthetics have significantly improved, the interface is cleaner, the SVG graphics generation capability is stronger, and the initial front-end output is more concise. In addition, there are significant improvements in game development. However, objectively speaking, Gemini 3.5 Pro is unlikely to challenge the leading position of Anthropic's Fable5 model in coding capabilities.
The image generation star returns, Nano Banana Pro aims at GPT-Image 2
Google's traditional advantage with the Gemini series has always been the breadth of world knowledge, a point that DeepSeek officially acknowledged as Google being the strongest. The higher the data quality, the better the model's overall performance. With the new foundation combined with better data, performance is expected to significantly improve. However, due to limitations in parameter scale, Gemini 3.5 Pro is unlikely to become the number one model immediately.
Google still holds another ace — image generation. Its self-developed Nano Banana image generation model once dominated the top spot globally for a long time, but recently it was surpassed by OpenAI's GPT-Image 2. This time, Google will launch the next-generation Nano Banana Pro large model based on Gemini 3.5 Pro, aiming directly to reclaim the throne in image generation and directly challenge GPT-Image 2. A dual competition covering text reasoning and image generation is about to begin in mid-July.