MSI’s U.S. store lists the 32GB Prestige N16 Flip AI+ C1XMT-067USY for $3,299.99 and labels it “Out of stock.” The listing gives October 23, 2026, as its release-and-ship date. The local-AI benchmark often associated with this laptop came from a different configuration: a 128GB review unit tested by Alex Ziskind.
What MSI lists for the U.S. base model
MSI’s U.S. listing for C1XMT-067USY specifies a 20-core NVIDIA Grace CPU, a 6,144-core Blackwell RTX GPU, 32GB of LPDDR5X unified memory, a 1TB SSD and Windows 11 Home. Its 16-inch OLED touchscreen has a 2,880 × 1,800 resolution and a 120Hz refresh rate.
Unified memory is shared by the processor and GPU. So the base model’s 32GB is the total shared pool, not 32GB of dedicated graphics memory. That distinction matters when comparing it with results from a higher-memory system running large AI models.
What Alex Ziskind’s 128GB test measured
Alex Ziskind tested a Prestige N16 Flip AI+ with 128GB of unified memory. In his llama.cpp test of GPT-OSS 120B, he reported generation rates of 65.4 tokens per second on the MSI laptop and 50.6 tokens per second on NVIDIA’s DGX Spark.
| System | GPT-OSS 120B generation rate in llama.cpp |
| MSI Prestige N16 Flip AI+ with 128GB | 65.4 tokens/s |
| NVIDIA DGX Spark | 50.6 tokens/s |
Tokens per second measures how quickly the system generates text in that test. The figures compare those two systems on that model and software; they are not results for every workload or for the 32GB base configuration.
NeoTeo’s earlier RTX Spark coverage looks at local-AI use from a separate angle.
What happened during a sustained workload
During a 20-minute test with the CPU and GPU under combined load, Ziskind reported an average chip temperature of about 199°F and a peak of about 209.5°F. He observed no significant thermal throttling in that run. Both the temperatures and the throttling result describe his tested unit and workload.
The U.S. base configuration has 32GB of shared memory; Ziskind’s local-model results came from a 128GB sample. That capacity difference is central to what those large-model test results say about the configuration MSI lists.
