Rent an NVIDIA Titan Xp GPU with 12 GB VRAM
Rent Titan Xp cloud GPU servers by the hour on Cloudj: 12 GB VRAM, NVIDIA Pascal and about 11.7 TFLOPS FP32 with 410 GB/s of memory bandwidth. It is the fastest Pascal consumer card in our list, a fit for hobbyists and researchers running FP32 work, quantized 7B–13B LLMs and SD 1.5 at a 160 W power limit. Check that your CUDA and framework versions still support this GPU, pick a live configuration and connect over SSH with your key.
Titan Xp specifications
VRAM per GPU
12 GB
Architecture
NVIDIA Pascal
Compute capability
6.1
FP32 compute
≈ 11.7 TFLOPS
Memory bandwidth
≈ 410 GB/s
Power limit
160 W
Values reported by the infrastructure provider for machines currently listed; real performance varies by machine.
Live Titan Xp configurations and prices
Out of stock right now
Browse available GPUsWhat you can run on the Titan Xp cloud GPU
Quantized 7B–13B LLMs
12 GB runs 8-bit 7B or 4-bit 13B models in Ollama, and 410 GB/s keeps generation usable.
FP32 training and research
11.7 TFLOPS FP32 handles classic deep learning and research code that runs in full precision.
SD 1.5 workflows
Run Stable Diffusion 1.5 and basic ComfyUI graphs; SDXL works with memory-saving settings.
Reproducing older papers
Rerun Pascal-era baselines on the hardware class they were originally trained on.
Is the Titan Xp the right GPU for your workload?
Choose the Titan Xp for budget-conscious development, legacy CUDA compatibility testing, or lightweight inference tasks requiring up to 12 GB of VRAM. Opt for a newer or larger GPU if your workload requires modern tensor cores or more memory capacity, or pick a smaller consumer card for minimal display and compute scripts.
Guides for Titan Xp
Titan Xp cloud GPU: frequently asked questions
Rent the Titan Xp cloud GPU by the hour
Pick a live configuration, choose a template and connect over SSH with your key. Billing stops when you destroy the instance.