Affiliate links on Android Authority may earn us a commission. Learn more.
Google Tensor G3: What to expect from the Pixel 8 processor

Although the Google Pixel 7a has not long hit store shelves, we’re almost set for the launch of Mountain View’s next-gen flagship smartphone — the Google Pixel 8. There will undoubtedly be exciting new features aplenty, whether it’s the new camera or software tweaks, but powering them all will be the hotly anticipated Tensor G3 processor.
Google is keeping the successor to its Tensor G2 chip, which powers a range of the brand’s latest products like the Pixel Tablet, under wraps for now. But we can figure out a little bit about what the chipset will look like and when it will debut based on information currently circling the web.
Will there be a Google Tensor G3 processor, and when will it arrive?

With each flagship smartphone release dating back to 2021’s Pixel 6 series, Google has debuted a new Tensor processor. The upcoming Pixel 8 range is expected to follow suit. According to a leaked roadmap of upcoming products obtained by Android Authority, Google’s “zuma” chip (aka the Tensor G3) will debut in the Pixel 8 and Pixel 8 Pro.
Google has confirmed that the Pixel 8 series will launch on October 4, and we expect to hear details on the new Tensor G3 chip at the event.
If the Tensor G2 is anything to go by, the Tensor G3 could also power an extensive range of Google products throughout 2024. The G2 ended up in the 2023’s A series budget smartphone, the Pixel Tablet, and first-gen Pixel Fold, bringing the chip’s AI processing smarts to a variety of form factors. Refreshes of at least some of these products are quite likely in 2024, and we’ve already seen very early reports that the Pixel 8a may feature a toned-down Tensor G3.
What features will the Tensor G3 have?

Google’s Tensor has, historically, not followed the traditional cadence of CPU and GPU upgrades, so predicting exactly what components will go into its next flagship chipset is particularly difficult. However, some leaks and rumors can point us in the right direction.
An upgraded CPU
The first hint came from Mishaal Rahman, who spotted support for Advanced Memory Protection for future chips in the Android 14 developer preview. This feature requires the Memory Tagging Extension (MTE) supported by ARMV9 CPUs.
Arm debuted the ARMV9 architecture with its Cortex-X2, Cortex-A710, and Cortex-A510 CPUs in 2021, all of which would be an upgrade over the Cortex-X1, A78, A76, and A55 setup employed by the original Tensor and Tensor G2. In addition, the newer Arm Cortex-X3 and A-715, as well as upcoming next-gen parts, are all compatible with MTE too.
Theoretically, Google could use any of these off-the-shelf cores for the Tensor G3, though the brand has historically stuck to older and, therefore, cheaper components. However, the most recent rumors point to a 1x Cortex-X3, 4x A715, and 4x A510 CPU configuration for the G3. Thanks to a source inside Google and a Pixel 8 system dump, we believe the Tensor G3 CPU layout will look close to the table below.
Tensor G3 (zuma) | Tensor G2 (gs201) | Tensor (gs101) | |
---|---|---|---|
Big cores | Tensor G3 (zuma) 1x Cortex-X3 @ 3.0GHz | Tensor G2 (gs201) 2x Cortex-X1 @ 2.85GHz | Tensor (gs101) 2x Cortex-X1 @ 2.8GHz |
Mid cores | Tensor G3 (zuma) 4x Cortex-A715 @ 2.45GHz | Tensor G2 (gs201) 2x Cortex-A78 @ 2.3GHz | Tensor (gs101) 2x Cortex-A76 @ 2.25GHz |
Little cores | Tensor G3 (zuma) 4x Cortex-A510 @ 2.15GHz | Tensor G2 (gs201) 4x Cortex-A55 @ 1.8GHz | Tensor (gs101) 4x Cortex-A55 @ 1.8GHz |
If true, this would push performance much closer to and perhaps even marginally ahead of the current Snapdragon 8 Gen 2, at least in the CPU department. Such a move could also indicate a completely 64-bit-only design, which would be an Android first and would match Google’s push with 64-bit-only software for the Pixel 7.
Improved gaming performance

The latest Tensor G3 leak obtained by Android Authority also points to a move to an Arm Immortalis G715 GPU in a ten-core configuration. This would be a decent upgrade to the Tensor G2’s Arm Mali-G710 seven-core setup, which currently doesn’t compete with the best in the business for graphics grunt. Arm says there’s a 15% ISO-process performance gain and 2x machine learning improvement with the G715 compared to the G710, plus the extra cores would lead to a noticeable performance gain.
However, the ten-core setup would be slightly smaller than the 11-core version in the MediaTek Dimensity 9200. Ballparking the performance improvements also suggests that the Tensor G3 will come in behind the current Snapdragon 8 Gen 2 in gaming performance. However, it remains to be seen how that plays out regarding real-world games.
Tensor G3 (zuma) | Tensor G2 (gs201) | Tensor (gs101) | |
---|---|---|---|
GPU Core Model | Tensor G3 (zuma) Mali-G715 (Immortalis) | Tensor G2 (gs201) Mali-G710 | Tensor (gs101) Mali-G78 |
Core Count | Tensor G3 (zuma) 10 | Tensor G2 (gs201) 7 | Tensor (gs101) 20 |
Frequency (shaders) | Tensor G3 (zuma) 890MHz | Tensor G2 (gs201) 848MHz | Tensor (gs101) 848MHz |
Immortalis branding in the Tensor G3’s G715 GPU rumor indicates ray tracing support. While far from an essential feature (we’ve only just seen our first titles supporting it), this does suggest that serious mobile gamers may be a little more impressed by the Pixel 8 series than the current generation models. If it turns out to be true, of course.
Next-gen AI smarts

Machine learning capabilities are what have kept the Tensor range in contention with more conventionally powerful phones powered by the Snapdragon 8 Gen 2 and Apple’s Bionic A16. We expect Google to further improve the upcoming chip’s Tensor Processing Unit (TPU).
According to our Tensor G3 leak, the SoC features a new version of Google’s custom TPU. Codenamed “Rio,” it will run at 1.1GHz, up from 1.00GHz in the Tensor G2’s “Janerio” TPU, along with other unspecified improvements.
Unfortunately, Google keeps its TPU sauce a closely guarded secret. Still, the new machine learning core will likely improve the performance and efficiency of offline voice tasks, such as real-time language translation.
… and more

Further image signal processor improvements to HDR and object segmentation are also likely, keeping the Pixel 8 series in contention for one of the best camera phones available. A second-generation “GXP” digital signal processor is rumored to be in the works, which handles a range of common image processing steps and should offer further improvements over the previous generation.
According to other rumors, the Tensor G3 will be manufactured on Samsung’s 4nm production line. That’s an improvement over the older 5nm process used by the Tensor G2 and should help the chip run more efficiently (for better battery life) and cooler (for longer sustained performance). Furthermore, this is rumored to be the first chip built by Samsung to use FO-WLP packaging, further improving thermal performance.
However, Samsung 4nm and low yields were behind the overheating issues with the Snapdragon 8 Gen 1. So, it remains to be seen if Samsung can catch up or surpass TSMC’s N4 node employed by MediaTek and Qualcomm for their latest flagship chips. Based on the ongoing reshuffling at Samsung Semiconductor, it’s not a given that the Tensor G3 will be free from the thermal issues that plagued 2022 graduates from Samsung’s foundry. However, Samsung is now on its third 4nm iteration, which has apparently seen significant yield improvements that help close the cap on its manufacturing rival. Either way, a cooler chip is very high on our Tensor G3 wishlist.
Finally, leaks also list new support for the latest UFS4.0 memory standard, the first mobile platform to support AV1 video encoding, and a slightly revised Exynos Modem 5300 that’s found in the Pixel 7 series
What we want to see from the Google Tensor G3
Given the lack of CPU upgrades since the first-gen chipset, a leap to the Arm Cortex-X3 and company would go a long way to addressing concerns that the Tensor series lags behind the competition. A GPU upgrade to the latest gen would also help, with or without the inclusion of fledgling ray tracing support. But our biggest complaint with Tensor to date has been high temperatures and battery drain, which Google also seems likely to address by moving to a smaller manufacturing node and tweaking its modem setup.
If the above turns out to be true, Google will have checked pretty much everything off our Tensor G3 wishlist. In that case, there’s even more reason to be excited about the arrival of the Pixel 8 series later in the year.