Google launches Pixel 11 on Tensor G6 with on-device Gemini Intelligence at $899
The Pixel 11 moves Tensor to TSMC's 2nm process for the first time, sets 16 GB of RAM as the floor for on-device Gemini LLM workloads across the premium lineup, and starts at $899, up $100 from the Pixel 10. Google confirmed all three at its Made by Google 2026 event in New York City on August 12.
What
Tensor G6 breaks from the Samsung fab used in every prior Tensor generation and instead runs on TSMC's 2nm N2 process. The base Pixel 11 starts at $899 for 256 GB of storage; the 128 GB option is discontinued from the lineup (per Android Central's Made by Google 2026 live blog). Every premium Pixel 11 model ships with 16 GB of RAM, a baseline Google set to support running Gemini LLM workloads directly on the device.
Three AI features shipped with the hardware. Gemini Intelligence provides deeper cross-app integration running on-device through the Tensor G6. Proactive Assistance is a Pixel 11 exclusive: it surfaces contextual suggestions from apps without a user prompt, building on the Magic Cue system introduced with the Pixel 10 (per Android Authority's feature analysis). Rambler handles voice input, automatically stripping filler sounds and formatting the transcript before it lands on screen.
Google also unveiled the Pixel 11 Pro Fold at $1,899, the Pixel Watch 5, and a Pixel Tag tracker at the same event (per Android Authority's launch preview). The full lineup ships August 20.
Why it matters
The TSMC 2nm switch addresses Tensor's longest-running hardware critique. Prior Tensor chips drew consistent criticism for thermal throttling and battery drain relative to Qualcomm's Snapdragon parts. The new fab process, paired with a MediaTek M90 modem replacing the Samsung-derived modem used in prior Pixel generations, targets both problems in one hardware revision.
The 16 GB RAM floor changes what the Pixel 11 can do offline. At $899, running a local LLM on-device was not achievable on prior Tensor hardware. Locking 16 GB across every premium model signals that on-device inference is now a core product commitment rather than a high-end option. AI tool operators evaluating on-device AI should watch whether Google's local inference results hold up against Apple's A-series chips on comparable workloads.
Proactive Assistance raises a separate question for IT teams. The feature reads context across apps without a user trigger, which requires broad data access. Whether enterprise mobile device management platforms can restrict that access will determine how deployable Pixel 11 is in managed environments.
What to watch next
Independent benchmarks on Tensor G6 efficiency and thermal performance will surface within days of the August 20 ship date. Watch also whether Google pushes Proactive Assistance to Pixel 10 hardware via a software update, or holds it as a Pixel 11 incentive.
Sources
- Android Central: Made by Google 2026 live blog (primary)
- Android Authority: Proactive Assistance feature analysis (secondary)
- Android Authority: Pixel 11 launch preview (secondary)
