Test records for a development sample of AMD's "Gainsborough" APU, rumored to be intended for the Steam Deck 2, have been reported. The sample reportedly has a 6-core, 12-thread CPU and 24GB of memory. That is a 50% increase in CPU cores over the current Steam Deck's four, and the memory transfer rate reportedly reaches 9600 MT/s. However, these figures do not represent official specifications for the next Steam Deck, nor do they mean gaming performance will improve by 50%. Comparing them with the specs Valve has published for the Steam Deck LCD and OLED models shows which improvements can be read from the test records and which performance differences cannot yet be judged.
Records of a development sample with 6 cores and 24GB of memory surface
A TechPowerUp article dated October 20 reported on test results for Gainsborough that were said to have been posted to OpenBenchmarking. According to the report, the CPU has 6 cores and 12 threads with a 2.4GHz base clock. It uses 24GB of LPDDR5 memory with a transfer rate of 9600 MT/s. The core count is 50% higher than the current Steam Deck's four cores, giving us concrete information about the CPU and memory configuration of the chip believed to be the next-generation APU.
Notebookcheck examined the test records in detail and found that the result was named "0807gpufull" and was said to have been uploaded on August 7, 2026. The board name is recorded as "AMD GBR-FF6-SKT" and the graphics vBIOS as "113-GBRGENERIC-001." Notebookcheck judged this identifying information to be related to Gainsborough. However, this information alone is not enough to conclude that it is the APU Valve will use in the next Steam Deck.
What matters is distinguishing between the configuration of a development sample and the specifications of a handheld that actually goes on sale. Even if the test environment had 24GB of memory, the retail product will not necessarily have the same capacity, and the CPU core count alone cannot tell us the GPU specifications. According to Notebookcheck, AMD and Valve have not officially announced Gainsborough, and no products planned to use it have been confirmed. It should also be noted that TechPowerUp and Notebookcheck are covering the same OpenBenchmarking record, so their reports do not constitute independent verification.
CPU and memory capacity up 50%; transfer-rate gain varies by model
According to Valve's official specifications, both the Steam Deck LCD and OLED models use a 4-core, 8-thread CPU based on the Zen 2 architecture, and both have 16GB of memory. The memory transfer rate differs, however: 5500 MT/s for the LCD model and 6400 MT/s for the OLED model. When comparing against the figures reported for Gainsborough, the percentage increase therefore depends on which model is used as the baseline.
| Item | Steam Deck LCD (official specs) | Steam Deck OLED (official specs) | Gainsborough (reported) |
|---|---|---|---|
| CPU cores / threads | 4 cores / 8 threads | 4 cores / 8 threads | 6 cores / 12 threads |
| Memory capacity | 16GB | 16GB | 24GB |
| Memory transfer rate | 5500 MT/s | 6400 MT/s | 9600 MT/s |
| APU power | 4–15W | 4–15W | Unconfirmed |
The figures for the current models come from Valve's official specifications, while the Gainsborough figures are based on reports from TechPowerUp and Notebookcheck. CPU cores rise from 4 to 6 and memory capacity from 16GB to 24GB, both increases of 50%. The memory transfer rate is 50% higher than the OLED model's 6400 MT/s, but about 74.5% higher if the LCD model's 5500 MT/s is used as the baseline. Lumping everything together as "50% more than the current Steam Deck" overlooks the differences between the LCD and OLED specifications.
Memory capacity, transfer rate, and bandwidth are also different metrics. Capacity, expressed in GB, indicates how much data can be stored, while MT/s indicates the number of data transfers per second. Memory bandwidth, the amount of data that can be moved per unit of time, depends not only on the transfer rate but also on the memory bus width. Since Gainsborough's memory bus width has not been confirmed, it cannot be asserted that memory bandwidth has increased 1.5 times. All that can be read from this test record are the reported values for memory capacity and transfer rate.
FurMark score is close to the current Steam Deck, but rendering conditions differ
In the results of the GPU stress test FurMark presented by TechPowerUp, Gainsborough scored 3508 points, compared with 3323 for the Steam Deck used as a reference. A simple comparison of scores puts Gainsborough about 5.6% ahead, but the two were not measured under the same conditions. Gainsborough's rendering resolution was 1024×768, versus 1280×800 for the Steam Deck, and according to Notebookcheck, Gainsborough was tested in windowed mode.
Calculating the pixel counts from these resolutions gives 786,432 pixels for Gainsborough and 1,024,000 for the Steam Deck. In other words, Gainsborough was tested with about 23.2% fewer pixels. Because the rendering load on the GPU differs, a performance gap cannot be judged from the scores alone.
That said, it is also not appropriate to simply adjust the scores by pixel count to estimate performance at the same resolution. Rendering load is not determined by resolution alone, and it is not known whether the two systems' power settings, drivers, and other conditions matched. Furthermore, TechPowerUp refers to the comparison target only as the Steam Deck and does not say whether the LCD or OLED model was used. Based on this score difference, it is currently difficult either to judge Gainsborough's performance as disappointing or to predict that it will improve significantly because it is still in development.
In handhelds, performance per watt matters more than CPU core count
The Steam Deck OLED's official specifications list the APU's operating power as 4–15W. In a handheld, what matters is not just the CPU core count or peak performance, but how smoothly games can run within a limited power budget. However, the reported Gainsborough test records do not show whether it was operating under the same power conditions as the current Steam Deck. A 6-core, 12-thread configuration alone therefore cannot be used to assess improvements in performance or power efficiency over the current models.
Even if the CPU core count rises 50%, game performance will not necessarily improve by the same proportion. If games cannot make full use of the additional cores, the benefit will be limited. And if the performance constraint lies in the GPU or memory, adding CPU cores may not translate into higher frame rates. This does not deny that Gainsborough may offer a performance improvement; it means that an increase in CPU cores and an actual gain in game performance need to be evaluated separately.
Battery life is likewise hard to predict from the APU configuration or benchmark scores alone. Beyond performance when running real games under the same power conditions, verification of the complete product is needed, including the cooling system, chassis design, and battery capacity. While these test results offer clues about the configuration and behavior of a development sample, they do not show how much performance or battery life a finished handheld will deliver.
For Gainsborough to serve as a basis for judging the next Steam Deck's performance, Valve must first confirm its adoption. After that, comparisons with the current models in real games, with matched resolution, rendering settings, and power consumption, will be needed. The key question going forward is how far the reported increases in CPU cores and memory capacity translate into better gameplay experience and power efficiency.