When will Google DeepMind make Gemini 3.5 Deep Think mode publicly available, meaning any member of the public can access it (including through a paid consumer subscription tier such as Google AI Ultra), rather than it being limited to trusted testers or partner previews?

rationale:

Update August 8, 2026: Set against related questions, this was shifted slightly to align with a consistent 2-to-6 week safety testing lag following the anticipated base model launch .

Current Status and the 3.5 Pro Blocker

As of August 2026, Gemini 3.5 Deep Think is not publicly available, and its release is fundamentally gated by the delayed Gemini 3.5 Pro base model. Google announced the 3.5 family at I/O on May 19, 2026, targeting a June rollout for 3.5 Pro blog.google. That target was missed, with reports indicating the model fell short of internal coding performance goals 9to5google.com. When Google launched its latest batch of models on July 21, including 3.6 Flash and 3.5 Flash-Lite, 3.5 Pro was conspicuously absent and described as still in partner testing 3 sources. Consequently, current Deep Think documentation and public access via Google AI Ultra remain anchored to the Gemini 3.1 cycle deepmind.google.

Near-Term Catalysts and Market Expectations

Prediction markets currently price an imminent August launch for the base Pro model polymarket.com, and some leaks suggest an upgraded Deep Think mode could be a headline feature nokiapoweruser.com. However, these markets have already been wrong about consecutive launch targets in June and July. Furthermore, public availability under this bar specifically requires open consumer access, such as through a paid Google AI Ultra subscription support.google.com. Even if the base model drops soon, Deep Think has historically required additional evaluation, safety testing, and a staged rollout before hitting consumer tiers.

Historical Pro-to-Deep-Think Lag

The historical lag between a flagship Pro model and its maximum-reasoning counterpart varies significantly. The transition from Gemini 3 Pro to Gemini 3 Deep Think took roughly two and a half weeks to reach Ultra subscribers en.wikipedia.org. The Gemini 3.1 update was essentially concurrent deepmind.google, while the earlier 2.5 cycle took roughly four months blog.google. Balancing the pressure to deliver a delayed Ultra-tier differentiator against the reality of an underperforming base model, a late August or September Pro release likely places the median public Deep Think rollout in early November. This accounts for a 2- to 8-week lag for a stable reasoning mode to clear evaluations and reach subscribers.

Skip, Rebrand, and Execution Risks

The most substantial risk—and the driver of the long right tail into 2027—is that Google skips or rebrands the 3.5 Deep Think label entirely. Google's version naming is already decoupled, with 3.6 Flash actively shipping while the Pro/Deep Think line sits at 3.1 2 sources. During the Q2 earnings call, Sundar Pichai notably deflected 3.5 Pro timeline questions to highlight Gemini 4, which is currently in its most ambitious pre-training run yet 3 sources. If 3.5 Pro continues to slip or fails to justify an expensive parallel-reasoning harness, Google may simply bridge the gap until a Gemini 4 or 3.6 Pro release, leaving the 3.5 Deep Think criteria unresolved. While DeepMind occasionally retroactively versions its products deepmind.google, this skip/rebrand risk is material enough to push the 75th and 90th percentiles into H1 2027.