$3,820.00 cheaper than the new price!!
Management number | 202046137 | Release Date | 2025/09/19 | List Price | $3,820.00 | Model Number | 202046137 | ||
---|---|---|---|---|---|---|---|---|---|
Category |
The Lenovo 4X67A14926 is a PCIe graphics card that carries the NVIDIA Tesla T4 GPU alongside 16 GB of GDDR6 memory. Designed for servers in the ThinkSystem family, it uses a passive heatsink and a low-profile bracket, so it relies on the chassis airflow already present in data center racks rather than an onboard fan.
This card excels when a server needs extra horsepower for artificial intelligence inference, real-time video processing, or the delivery of virtual desktops. A team running edge analytics on a ThinkSystem SE350 at a remote site can insert the card to speed up deep-learning models without changing the server’s physical footprint. In a core data center, administrators can install multiple cards in an SR650 to support dense VDI sessions or to transcode large volumes of video content for streaming services.
Because the Tesla T4 architecture supports mixed-precision computing, workloads ranging from FP32 down to INT8 can be executed efficiently, letting customers balance performance and accuracy according to application needs. The PCIe 3.0 x16 interface ensures the GPU has sufficient bandwidth to move data between system memory and the accelerator, while the single-slot, low-profile format leaves more space and power headroom for additional expansion cards in the same server.
Before deployment, it is important to verify that the target system provides adequate front-to-back airflow, as the card’s passive design counts on that airflow for thermal management. When this requirement is met, the Lenovo 4X67A14926 offers a straightforward way to add GPU acceleration to existing ThinkSystem infrastructure without major changes to power or cooling budgets.
Highlights:If you notice any omissions or errors in the product information on this page, please use the correction request form below.
Correction Request Form