NVIDIA’s RTX Spark is shaping up to be one of the most interesting laptop chips of the year, but a new Cinebench 2026 leak suggests its CPU performance may not be as impressive as many expected. The chip, reportedly tested inside Microsoft’s Surface Laptop Ultra, uses a 20-core CPU configuration built around older Cortex-X925 and Cortex-A725 cores. While the single-core result looks respectable, the multi-core score appears to fall behind rival high-end laptop processors.
According to the leaked Cinebench 2026 numbers, RTX Spark scored 540 points in single-core and 5,771 points in multi-core performance. That single-threaded result puts it in a competitive position against Apple’s M3 Max, showing that NVIDIA’s custom-tuned Cortex cores can still deliver solid performance in lighter workloads. However, the multi-core score tells a different story.
Compared with Apple’s newer M4 Max, RTX Spark reportedly trails by a significant margin in multi-threaded workloads. The M4 Max is said to be around 35.6 percent faster in Cinebench 2026 multi-core testing, which is a notable gap for a chip expected to compete in premium Windows laptops.
The leak also highlights an important detail about the Surface Laptop Ultra’s performance settings. High Performance mode is not enabled by default, meaning users may not see the chip’s full potential unless they access a hidden configuration. In this test, the setting was reportedly enabled, allowing RTX Spark’s Cortex cores to draw close to 50W while running at around 2.80GHz.
That raises some questions. Cortex-X925 cores can boost up to 4.00GHz under ideal thermal conditions, so the lower observed frequency may point to software limits, power restrictions, or cooling constraints inside the Surface Laptop Ultra. If the laptop’s thermal design cannot properly handle the chip at higher sustained power levels, RTX Spark may struggle to show its best CPU performance in real-world workloads.
Previous reports suggested that NVIDIA modified the Cortex-X925 cores to better suit PC-class tasks, with the goal of maintaining higher clock speeds under demanding workloads. Even with those adjustments, RTX Spark appears to face a tough battle on the CPU side, especially against Apple’s latest laptop silicon.
Still, the full picture is not entirely negative. The RTX Spark’s biggest strength may not be its CPU at all, but its Blackwell-based GPU and unified memory design. This combination could give NVIDIA a major advantage in graphics-heavy workloads, AI applications, and gaming.
The chip is expected to handle demanding titles such as PRAGMATA and Alan Wake 2 with path tracing enabled, while also being capable of running large language models with up to 120 billion parameters. That makes RTX Spark especially interesting for creators, developers, gamers, and AI users who care more about GPU performance than traditional CPU benchmarks.
This is where NVIDIA could separate itself from Apple and Qualcomm. While the leaked Cinebench 2026 results suggest RTX Spark may lag in multi-core CPU performance, its GPU capabilities could make it one of the most powerful laptop platforms for gaming and AI acceleration.
For now, the RTX Spark remains a mixed story. Its single-core performance is solid, its multi-core performance appears underwhelming, and its graphics potential looks extremely promising. The real test will come when commercial laptops powered by RTX Spark begin shipping and reviewers can evaluate performance, thermals, battery life, and gaming results across multiple devices.
If NVIDIA can address the CPU limitations in a future successor while keeping its GPU advantage intact, RTX Spark could become a serious challenger in the high-performance laptop chip market.
