# We tested the Pixel 11’s Tensor G6 chip, and the results are brutal

> Source: <https://www.androidauthority.com/tensor-g6-benchmarks-tests-3699714/>
> Published: 2026-08-20 09:00:22+00:00

Affiliate links on Android Authority may earn us a commission. [Learn more.](https://www.androidauthority.com/external-links/)

# We tested the Pixel 11's Tensor G6 chip, and the results are brutal

Aug 20, 2026 — 5:00 AM ET

I’ll be frank, my hopes for the [Tensor G6](https://www.androidauthority.com/tensor-g6-deep-dive-3695968/) were mixed coming into the launch of the [Pixel 11 series](https://www.androidauthority.com/google-pixel-11-hands-on-impressions-3699583/). Google has made meaningful upgrades to the CPU and TPU for traditional app and AI workloads, respectively, while switching to what should be a more efficient modem. But pairing the [Pixel 11 Pro with just 12GB of RAM](https://www.androidauthority.com/pixel-11-pro-not-for-android-enthusiasts-3697917/), down from 16GB last year, undermines its high-end AI credentials. Meanwhile, Google has dropped a CPU core from its configuration, reducing the benefits of its more modern architecture.

While we can’t blame Google for the current state of the global RAM supply chain, its decision to switch to yet another mid-range graphics chip is harder to defend. Still, now that we have the phone in hand (and managed to get around Google blocking benchmarks via the Play Store), we’ve been able to run our usual suite of benchmarks to get a feel for where the Pixel 11 series falls in the performance pecking order.

The results aren’t great.

### What do you think of Google's Tensor G6 approach?

## Yearly improvements, but the pace is too slow

Google Tensor G6 | Google Tensor G5 | Google Tensor G4 | Google Tensor G3 | |
|---|---|---|---|---|
CPU | Google Tensor G61x Arm C1-Ultra (4.11GHz) 4x Arm C1-Pro (3.38GHz) 2x Arm C1-Pro (2.65GHz) | Google Tensor G51x Arm Cortex-X4 (3.78GHz) 5x Arm Cortex-A725 (3.05GHz) 2x Arm Cortex-A520 (2.25GHz) | Google Tensor G41x Arm Cortex-X4 (3.1GHz) 3x Arm Cortex-A720 (2.6GHz) 4x Arm Cortex-A520 (1.92GHz) | Google Tensor G31x Arm Cortex-X3 (2.91GHz) 4x Arm Cortex-A715 (2.37GHz) 4x Arm Cortex-A510 (1.70GHz) |
GPU | Google Tensor G6PowerVR CXTP-48-1536 (1.29GHz) | Google Tensor G5PowerVR DXT-48-1536 (1.1GHz) | Google Tensor G4Arm Mali-G715 (MC7?) (940MHz) | Google Tensor G3Arm Mali-G715 MC7 (890MHz) |
RAM | Google Tensor G6LPDDR5X | Google Tensor G5LPDDR5X | Google Tensor G4LPDDR5X | Google Tensor G3LPDDR5X |
Machine Learning | Google Tensor G6Fifth-gen Tensor Processing Unit | Google Tensor G5Fourth-gen Tensor Processing Unit | Google Tensor G4Third-gen Tensor Processing Unit | Google Tensor G3Third-gen Tensor Processing Unit |
Security Co-processor | Google Tensor G6Titan M3 | Google Tensor G5Titan M2 | Google Tensor G4Titan M2 | Google Tensor G3Titan M2 |
Modem | Google Tensor G6MediaTek M90 | Google Tensor G5Samsung Exnos | Google Tensor G4Samsung Exnos | Google Tensor G3Samsung Exnos |
Process | Google Tensor G6TSMC 3nm | Google Tensor G5TSMC 3nm | Google Tensor G4Samsung 4nm | Google Tensor G3Samsung 4nm |

To start our dive into the new Tensor G6, let’s look at performance over the last few years to help frame how the Pixel series has improved. Starting on the CPU side, the Tensor G6 moves up to an [Arm C1-Ultra](https://www.androidauthority.com/arm-c1-cpu-mali-g1-gpu-deep-dive-3595933/) at 4.11GHz for the big core, four Arm C1-Pro cores at 3.38GHz for multitasking power, and two Arm C1-Pro cores at 2.65GHz for energy efficiency. While that’s one less core than last year, the Tensor G6 has skipped ahead two CPU generations this time, which, in theory, should lead to a rather significant performance jump compared to the Tensor G5. Unfortunately, the benchmark results are more of a mixed bag.

According to Geekbench 6, single-core performance is up by around 18%, and multi-core performance gains are in the region of 21%. That’s an undoubtedly welcome boost and a very promising generational gain. The chip is certainly fast enough to ensure the Pixel 11 series feels very responsive for everyday and even more demanding tasks. However, if you were hoping for performance approaching that of [Samsung’s Exynos 2600](https://www.androidauthority.com/exynos-vs-snapdragon-galaxy-s26-benchmarks-3653459/), you may be disappointed.

The Exynos uses similar CPU cores at even more conservative clock speeds, yet its larger core count gives it a substantial advantage. My Galaxy S26 Plus scores 18% and 51% higher in single- and multi-core tests, respectively, than the Pixel 11 Pro XL. There are a couple of possible explanations for Google’s performance deficit: more aggressive thermal management, or perhaps smaller cache sizes. We know Arm’s powerhouse cores love big caches, but Google appears to be trying to save as much silicon area as possible.

But it’s the graphics situation that reveals Google’s stark lack of progress in one of mobile’s most popular commuting use cases. According to 3DMark’s Wild Life and Wild Life Extreme tests, peak performance has increased by about 7% to 10% compared to last year’s processor, depending on the exact test. Again, that’s a welcome boost and could be enough to turn somewhat wobbly frame rates into a mostly locked 60 fps. But it isn’t enough to transform the Pixel 11 into a top-performing gaming phone capable of sustaining 120 fps in the latest titles.

CPU, GPU, and TPU performance are up year-on-year.

Unfortunately, that’s about as good as the news gets for prospective Pixel gamers. Looking at the stress-test results suggests that over longer gaming sessions, particularly in warm climates, there may not be much to tell between the Pixel 11 and Pixel 10 series after just a few minutes of intensive gameplay. Performance actually dips to below last year’s GPU after just a few minutes of 3DMark’s stress test, indicating that those higher GPU clocks aren’t particularly sustainable in the long term. Still, real games often take longer to heat up a handset, and we see rival flagships drop performance far more quickly.

## Rival flagships offer far more performance

The real problem is that Google’s latest gains haven’t meaningfully closed the gap. Snapdragon-powered rivals retain a colossal 37% single-core and 52% multi-core advantage over the Pixel 11, and the next Snapdragon refresh is just a few months away. Likewise, Apple’s custom CPU cores offer a hefty 44% advantage in this test, while the deca-core Samsung Exynos 2600 smashes ahead to a 50% multi-core lead.

Google’s Tensor G6 is plenty powerful for daily tasks; its CPU setup scores about as well as the 2025-era Dimensity 9400, which is a very potent chip. However, CPU performance remains a very important metric, despite the increasing focus on features like AI and gaming. Not only does a powerhouse CPU ensure that apps launch quickly, videos encode promptly, and games run smoothly, but its main cores also serve as an important backup for running AI tasks that don’t yet tap into Google’s proprietary TPU hardware. As rival CPUs become faster while adding their own [SME2-class AI acceleration](https://www.androidauthority.com/exynos-vs-snapdragon-galaxy-s26-3641995/), Google’s decision to prioritize TPU performance at the expense of CPU and GPU muscle starts to look increasingly difficult to justify.

For gamers, Google’s AI-first strategy leaves the Pixel playing in the second league for another year.

In terms of peak performance, the Tensor G6’s PowerVR CXT graphics setup is easily bested. The Snapdragon 8 Elite Gen 5’s Adreno 840 produces more than double the score in 3DMark’s Wild Life Extreme test, as does the Arm G1-Ultra inside the Dimensity 9500 (+135% versus Tensor). Apple’s own silicon has a 60% advantage in this test, and the AMD Xclipse 940 in Samsung’s Exynos also bests Tensor by a notable 90%, almost doubling its peak performance. In other words, rival flagship phones selling for a comparable price can often offer more than double the frame rate or let you play games with far higher graphics settings than the Pixel 11.

Rivals more than double Tensor's graphics capabilities.

If there’s a silver lining for the Tensor G6, it’s that its stress-test performance means it gives up far less of its peak potential than these faster competitors. Google’s chip retains about 79% of its peak potential at the end of a 20-run stress test, compared to results ranging from 75% to just 50% for its rivals. However, other chipsets are so far ahead in raw performance that even significant temperature buildup and gameplay times leave them comfortably ahead of Google’s latest silicon, even under the worst conditions.

In addition, there’s still no ray-tracing support in the Tensor G6. While still not a mainstream mobile gaming technology, an increasing number of games are using it. The fact that other silicon vendors are now in their third or fourth generation of capable hardware, with Google not making a move of its own, highlights where its priorities lie.

## What is Tensor for?

Google’s vision for Tensor has always extended beyond traditional performance metrics. The fact that the company has focused on boosting TPU performance for on-device AI capabilities by a significant 50% this year, while CPU and GPU see just a fraction of that performance uplift, only serves to reinforce the point that Google is fully on board with the belief that new AI features will keep driving Pixel sales. Everyone else can fight it out for legacy benchmark titles as far as it’s concerned.

Fair enough. But it’s tough to ignore just how far behind Google is falling. Whether Google believes it or not, that has repercussions for a significant number of consumers. As discussed, CPU performance remains the backbone of increasingly demanding mobile experiences, from heavy content creation to [PC-class capabilities](https://www.androidauthority.com/android-desktop-mode-getting-started-3646933/). Likewise, gaming is a well-trodden mobile pastime that’s more popular than ever, and a good portion of consumers expect their increasingly expensive flagship phones to play the latest games and emulated classics at buttery-smooth frame rates.

There’s also futureproofing to consider. Google markets its handsets with seven-year lifespans yet ships phones with performance that’s easily a year, if not two, behind what rivals offer. Even Google’s own rate of change in the AI department has left previous-generation phones without some of its latest and greatest features, leaving us to question whether jumping into Pixel for the chip’s TPU prowess is really all that justifiable in the first place. Combined with this year’s RAM downgrade, Google’s AI-first pitch becomes difficult to square with the Pixel 11’s long-term prospects.

Perhaps the Tensor G6 offers a significant improvement in battery life that will put it in a favorable light; we’ll be testing that soon enough. But in the meantime, this year’s Tensor G6 deep dive leads to the same conclusion as the past few years: buy a Pixel 11 if you love Google’s unique features, but it might be best to skip it if you want the best performance money can buy.

Thank you for being part of our community. Read our [Comment Policy](https://www.androidauthority.com/android-authority-comment-policy/) before posting.