← Back to GPUs
NVIDIA Tesla P40

NVIDIA · Data Center

NVIDIA Tesla P40

$300$5000 MSRP

The NVIDIA Tesla P40 is the ultimate budget AI card — 24GB of VRAM for around on the used market. Based on the older Pascal architecture (2016), it lacks modern tensor cores and FP16 acceleration, making inference significantly slower than newer cards. But for hobbyists who want to experiment with 32B models at Q4 quantization without spending thousands, nothing else comes close on price. Requires a second GPU for display output and runs with a loud blower cooler.

Best ForCheapest 24GB VRAM card available — the budget AI experimenter's pick
Verdict24GB for is unbeatable value if you can tolerate the limitations.
AI
5/10
Gaming
1/10

Specifications

VRAM24GB GDDR5X
Memory Bandwidth346 GB/s
CUDA Cores3,840
Boost Clock1531 MHz
TDP250W
Power Connector1x 8-pin
Length267mm
Form FactorDual Slot
Release Year2016

AI Capabilities

Sweet Spot24GB VRAM

The professional standard. Handles most models with smart quantization.

Can run (Q4 quantized)

Llama 3.1 8BQwen 2.5 32BQwen 2.5 14BMistral 7BFLUX.1 DevStable Diffusion XLStable Diffusion 3.5 LargeHunyuanVideoCogVideoX-5BMochi 1LTX VideoStable Video DiffusionWan Video 14BCodestral 22BQwen 2.5 Coder 32BLLaVA 1.6 34BAlphaFold 2ESMFold (ESM-2 15B)ESM-2 3BscGPTRFdiffusionFine-tune Llama 8BTrain SDXL LoRATrain FLUX LoRA

Recommended system RAM for AI: 48GB+ (2x GPU VRAM for model overflow)

Pros

  • +24GB VRAM for $300 used — cheapest way to get 24GB
  • +Runs 32B models at Q4
  • +Great for inference experimentation

Cons

  • -No display output — needs a second GPU for video
  • -Old Pascal architecture — no FP16 tensor cores
  • -Very slow by modern standards
  • -Loud blower cooler
aibudget

Where to Buy