The long wait for Google’s next heavy hitter might almost be over. If the latest buzz and community leaks are anything to go by, Google is gearing up to release Gemini 3.5 Pro on August 12th.
While Gemini 3.5 Flash gave us a solid taste of what Google has been cooking, power users and developers have been holding out for the “Pro” build. Now, after weeks of subtle platform tests and benchmark spottings, it looks like the wait is paying off—and early feedback points to a massive step up in speed, reasoning, and raw coding power.
Here is a look at what has surfaced so far.
What’s Leaking Ahead of Launch?
Quiet Partner Testing & Benchmark Cameos
Google hasn’t officially announced the date, but they’ve clearly been putting the model through its paces. Enterprise partners have already been quietly testing early builds, and eagle-eyed users have spotted mysterious contenders over on LM Arena.
In blind side-by-side tests, these unnamed models have been consistently blowing existing benchmarks out of the water, especially when asked to handle strict instructions and complex, multi-layered logic.
That Insane 2M Token Context Window
One of the biggest headlines in the leak is a confirmed 2-million-token context window.
To put that into practical terms: you’ll be able to dump entire codebases, massive PDF manuals, or hours of high-res video into a single prompt without the model losing its train of thought halfway through.
A Much Smarter “Deep Think” Mode
Google is also doubling down on multi-step reasoning with an updated Deep Think mode. Instead of rushing to output the first plausible-sounding answer, Gemini 3.5 Pro takes a second to break down complex queries, sanity-check its own logic, and work through complex math or technical problems with far fewer hallucinations.
Serious Upgrades to Coding & Agent Workflows
If you use LLMs for software engineering, this is where things get really interesting. Early feedback from the A/B tests suggests a huge jump in real-world performance:
Cleaner, smarter code: It’s significantly better at managing multi-file projects, refactoring legacy scripts, and catching obscure bugs.
Better autonomy: Tool use and agentic workflows feel much smoother, making it far more reliable when plugged into CLI tools or complex automation pipelines.
The Verdict So Far
From what we’re seeing in the early A/B tests, Gemini 3.5 Pro isn’t just a minor update—it feels like a genuine leap forward. The boost in coding fluidity alone is going to make it a huge competitor in the current AI space.
For more of the latest tech breakdowns, device updates, and insider leaks, head over to NokiaPowerUser.
Are you planning to put Gemini 3.5 Pro through its paces when it drops, or are you waiting to see how the benchmarks hold up first?
