AMD’s next-generation Medusa Point APU lineup, based on the upcoming Zen 6 CPU architecture, has surfaced again in the Geekbench database—and this time the leak paints a much more exciting picture. Compared with the earlier appearance, the new listing suggests a major jump in benchmark performance and also reveals a notable new instruction feature that could matter for modern AI and productivity workloads.
The newly spotted chip is once again identified as a Ryzen 9-branded Medusa Point APU. According to the listing, it features 10 CPU cores and 20 threads, built on Zen 6. The core layout is shown as a 4+6 configuration, which points to a hybrid-style design approach that mixes classic high-performance cores with more efficiency-optimized cores.
Clock speeds in the leak are also interesting. The processor is listed with a 2.40 GHz base frequency, but the benchmark run appears to have it operating closer to 2.0–2.1 GHz during the test. Cache details include 32 MB of L3 cache and 10 MB of L2 cache—figures that help frame how AMD may be balancing latency, efficiency, and throughput for a mobile-focused APU in its target power class.
One of the biggest reveals in this new leak is instruction support. For the first time, Geekbench is reporting FP16 “AVX-VNNI” capability tied to Zen 6. In practical terms, that hints at full FP16 support in future Zen 6-based SoCs like Medusa Point, which could improve performance in workloads that benefit from FP16 operations—often including newer AI-accelerated tasks and certain content creation or compute-heavy applications optimized for these instructions.
On the performance side, the leaked Medusa Point entry scored 2300 points in Geekbench single-core and 13,002 points in multi-core. When compared against the average Geekbench results for the Ryzen AI 9 365 (a Strix Point Zen 5 APU), the Zen 5 chip comes out about 8% faster in single-core. However, the leaked Zen 6 Medusa Point result appears to pull ahead in multi-threaded performance by roughly 4.4%.
That comparison becomes even more notable when you consider operating clocks. The Ryzen AI 9 365 can boost up to around 5 GHz, while the Medusa Point sample in this test ran closer to the low 2 GHz range. Even allowing for normal variation between benchmark runs and systems, it’s an encouraging early signal for Zen 6 efficiency and scaling—especially if future silicon ends up reaching higher clocks or ships with additional tuning.
While early benchmarks should always be treated cautiously, the direction is clear: Zen 6 is shaping up as a meaningful step forward. With expectations of higher core counts, faster clocks, expanded cache options, and new platform features, Medusa Point could become an attractive upgrade target for users who keep their laptops longer and want a bigger generational leap when they finally move.
As of now, Medusa Point APUs are said to be targeting a 2027 release window, with more details expected around CES 2027 as AMD’s roadmap becomes clearer.
