A 57% multi-core jump. An 89% GPU leap. And an NPU that went from 44 TOPS to 200 TOPS in one shot. Xiaomi didn’t call this chip the O3 just to sound cool — the numbers back it up. According to Gizmochina, the Xring O3 is a genuine multi-generational upgrade over the Xring O1, and the comparison between these two chips is honestly kind of wild.
Both chips are built on TSMC’s 3nm node, but that’s where the surface-level similarities end. The O3 uses TSMC’s N3P process instead of the N3E used on the O1. That’s a subtle but meaningful difference — N3P squeezes out better performance and efficiency from the same general manufacturing generation. Xiaomi also packed 24 billion transistors into the O3, up from 19 billion, which pushed the die size from 114.7 sq mm to 133 sq mm. More transistors, more room to do more things.
The CPU story is where things get really interesting. The Xring O1 used a mix of ARM’s Cortex-X925, A725, and A520 cores — a fairly standard big-middle-little layout. The Xring O3 throws all of that out. It runs a full all-big-core design using ARM’s newer C1-series cores across all 10 cores. Peak clock speed goes up to 4.35 GHz, compared to 3.9 GHz on the O1. In Geekbench 6, that translates to a single-core score of 3,699 versus the O1’s 2,910, and a multi-core score of 14,350 versus 9,153. That multi-core result reportedly tops every other mobile chip right now, including the Apple A19 Pro and Snapdragon 8 Elite Gen 5.
The GPU upgrade is even more dramatic. The O3 uses a 16-core Mali-G2 Ultra NX, which is the first time this GPU has appeared in a mobile chip. Eight of those cores handle standard graphics work, and eight are dedicated neural network units. Those NX cores let games render at a lower resolution and then use a trained model to upscale the image and generate extra frames. It’s similar to what Nvidia does with DLSS on PC, and it’s a smart way to get better-looking games without just brute-forcing the GPU. Peak GPU frequency also jumped 25% to 1.6 GHz. In the 3DMark Wild Life Extreme test, the O3 scored 11,378 versus the O1’s 5,998. That’s an 89% improvement.
Memory is another first. The Xring O3 is the first mobile chip to use LPDDR6 RAM, running at up to 5333 MHz. That gives it a peak memory bandwidth of 113.8 GB/s, which is 48% more than the LPDDR5T in the O1. Faster memory means the CPU and GPU spend less time waiting around for data, and that affects everything from app loading to gaming to AI tasks.
Speaking of AI, the NPU is arguably the biggest leap of all. The Xring O1 managed 44 TOPS. The O3 does 200 TOPS. That’s not an incremental upgrade — that’s a completely different class of on-device AI capability. We’re talking about running complex AI assistants, real-time photo processing, and other heavy workloads entirely on the device, no cloud required. That matters a lot as smartphone makers push harder into AI features that users actually care about.
Here’s a quick look at the key spec differences side by side:
- CPU: All-big-core C1-series (O3) vs mixed Cortex-X925/A725/A520 (O1)
- Peak CPU speed: 4.35 GHz (O3) vs 3.9 GHz (O1)
- GPU: Mali-G2 Ultra NX with neural network units (O3) vs Mali-G925 Immortalis (O1)
- NPU: 200 TOPS (O3) vs 44 TOPS (O1)
- Memory: LPDDR6 at 5333 MHz (O3) vs LPDDR5T at 4800 MHz (O1)
- Transistors: 24 billion (O3) vs 19 billion (O1)
- Bluetooth: 6.0 (O3) vs 5.4 (O1)
So why does any of this matter beyond spec sheet bragging? Xiaomi is clearly serious about building its own silicon long-term. The original Xring O1 was already a strong debut that showed Xiaomi could compete with Qualcomm and MediaTek. The O3 is the company proving it can iterate fast and push the envelope on multiple fronts at once. That’s not easy, and it puts pressure on the rest of the industry.
The O3 is currently powering the Xiaomi Pad 9 Pro Max, with more devices expected to follow. If you’re in the market for a Xiaomi device and raw performance matters to you, this chip is the one to look for.