readygpu

NVIDIA RTX PRO 4500 Blackwell vs NVIDIA Tesla P40

For running models locally, priced in the UK. At 128k context the first runs 50 models fully in memory, the second 31.

In short

The NVIDIA RTX PRO 4500 Blackwell has more memory for models: 32 GB against 24 GB. At 128k context the NVIDIA RTX PRO 4500 Blackwell runs 50 of the models we track fully in memory and the NVIDIA Tesla P40 31; only the NVIDIA RTX PRO 4500 Blackwell holds Qwen 3.8 27B. On the models both run fully in memory, the NVIDIA RTX PRO 4500 Blackwell is about 2.6× as fast as the NVIDIA Tesla P40 (estimated, median across models).

Updated . Speeds are estimates; how they are worked out.

NVIDIA RTX PRO 4500 Blackwell

Not priced here

Add to build
NVIDIA Tesla P40

£234

eBay

Add to build

Specifications

SpecNVIDIA RTX PRO 4500 BlackwellNVIDIA Tesla P40
GPURTX PRO 4500 BlackwellTesla P40
Memory32 GB GDDR724 GB GDDR5
Memory bus256-bit384-bit
Memory bandwidth896 GB/s346 GB/s
FP16 compute54.94 TFLOPS11.8 TFLOPS
SoftwareCUDACUDA
Board power200 W250 W
PCIe generationGen 5Gen 3
PCIe lanesx16x16

What each runs

Estimated decode speed with the most precise quant that fits each one est.

ModelNVIDIA RTX PRO 4500 BlackwellNVIDIA Tesla P40
Qwen 3.8 27B18tok/sUD-Q6_Koffload
Qwen 3.6 27B18tok/sQ6_Koffload
Gemma 4 31B19tok/sQ4_K_Moffload
Qwen 3.6 35B-A3B64tok/sUD-Q6_Koffload
Qwen 3.5 27B18tok/sQ6_Koffload
Gemma 4 26B-A4B47tok/sQ8_022tok/sQ5_K_M
Qwen 3.5 35B-A3B65tok/sQ6_Koffload
GLM 4.7 Flash30tok/sQ6_Koffload
Nemotron 3.5 Lightning 30B-A3B90tok/sUD-Q5_K_Moffload
Kimi Linear 48B-A3B114tok/sQ4_K_Moffload
gpt-oss 20B59tok/sMXFP423tok/sMXFP4
Qwen3 Coder 30B-A3B23tok/sQ4_K_Moffload
Gemma 4 12B37tok/sQ8_014tok/sQ8_0
Qwen 3.5 9B40tok/sQ8_016tok/sQ8_0
Gemma 3 27B18tok/sQ5_K_Moffload
Ministral 3 14B18tok/sQ5_K_Moffload
Qwen3 14B17tok/sQ5_K_Moffload
DeepSeek R1 0528 Qwen3 8B20tok/sQ8_0offload
Qwen3 8B20tok/sQ8_0offload
Qwen 3.5 4B63tok/sQ8_024tok/sQ8_0
Ministral 3 8B20tok/sQ8_09tok/sQ4_K_M
Granite 4.2 8B18tok/sQ8_0offload
Gemma 4 E4B46tok/sQ8_018tok/sQ8_0
LFM2.5 8B-A1B44tok/sQ8_017tok/sQ8_0
Llama 3.1 8B23tok/sQ6_K9tok/sQ5_K_M
Qwen2.5 0.5B259tok/sQ8_0100tok/sQ8_0
Qwen3.5 0.8B228tok/sQ8_088tok/sQ8_0
Qwen3.5 2B181tok/sQ5_K_M70tok/sQ5_K_M
DeepSeek R1 Distill Qwen 1.5B98tok/sQ8_038tok/sQ8_0
Mellum2.1 12B A2.5B Thinking73tok/sQ8_028tok/sQ8_0
Spark X2.5 4B70tok/smlx-4bit27tok/smlx-4bit
MiniCPM5 2B67tok/sQ8_026tok/sQ8_0
FrogNano 4B 260962tok/sQ8_024tok/sQ8_0
LFM2.5 2.6B51tok/sQ8_020tok/sQ8_0
LFM2.5 VL 3B51tok/sQ8_020tok/sQ8_0
Kimi VL A3B Thinking 250646tok/sQ8_018tok/sQ8_0
Kimi VL A3B45tok/smlx-8bit18tok/smlx-8bit
GLM 4 9B 041441tok/smlx-5bit16tok/smlx-5bit
granite vision 4.1 4b39tok/sQ8_015tok/sQ8_0
granite 4.2 3b38tok/sQ8_015tok/sQ8_0
GLM 4.6V Flash36tok/sQ8_014tok/sQ8_0
DeepSeek R1 Distill Qwen 7B36tok/sQ8_014tok/sQ8_0
GLM 4.1V 9B Thinking32tok/smlx-8bit13tok/smlx-8bit
Ministral 3 3B Instruct 251232tok/sQ8_012tok/sQ8_0
LFM2.5 24B A2B27tok/sQ6_Koffload
Muse Glimmer 30B22tok/sQ6_K10tok/sQ5_K_M
DeepSeek R1 Distill Llama 8B22tok/sQ8_09tok/sQ5_K_M
Ministral 3 8B Reasoning 251220tok/sQ8_09tok/sQ4_K_M
Qwen3.8 27B OBLITERATED19tok/smlx-4bitoffload
Ministral 3 14B Reasoning 251218tok/sQ5_K_Moffload
Qwen 3.5 122B-A10Boffloadoffload
Qwen3 Coder Nextoffloadwon't run
Nemotron 3 Super 120B-A12Boffloadwon't run
Mistral Small 4 119Boffloadoffload
Qwen3 Next 80B-A3Boffloadwon't run
Devstral Small 2 24Boffloadoffload
Qwen3 32Boffloadwon't run
Mistral Small 3.2 24Boffloadoffload
DeepSeek R1 Distill Qwen 32Boffloadwon't run
Phi-4 14Boffloadoffload
DeepSeek R1 Distill Qwen 14Boffloadoffload
Devstral Small 2507offloadoffload
Magistral Small 2506offloadoffload
Magistral Small 2507offloadoffload
Magistral Small 2509offloadoffload
granite 4.2 30boffloadwon't run
DeepSeek V4 Flashwon't runwon't run
GLM 5.3 Flashwon't runwon't run
Qwen 3.8 Flash Nextwon't runwon't run
MiniMax M2.7won't runwon't run
gpt-oss 120Bwon't runwon't run
GLM 4.5 Airwon't runwon't run
Llama 3.3 70Bwon't runwon't run
DeepSeek R1won't runwon't run
DeepSeek R1 0528won't runwon't run
DeepSeek R1 Distill Llama 70Bwon't runwon't run
DeepSeek R1 Zerowon't runwon't run
DeepSeek V3won't runwon't run
DeepSeek V3 0324won't runwon't run
DeepSeek V3.1won't runwon't run
DeepSeek V3.1 Terminuswon't runwon't run
DeepSeek V3.2won't runwon't run
DeepSeek V4 Flash 0731won't runwon't run
DeepSeek V4 Flash Vision Expwon't runwon't run
Devstral 2 123B Instruct 2512won't runwon't run
GLM 4.5won't runwon't run
GLM 4.5Vwon't runwon't run
GLM 4.6won't runwon't run
GLM 4.6Vwon't runwon't run
GLM 4.7won't runwon't run
GLM 5won't runwon't run
GLM 5.1won't runwon't run
GLM 5.2won't runwon't run
GLM 5.3won't runwon't run
Kimi Dev 72Bwon't runwon't run
Kimi K2won't runwon't run
Kimi K2 Instruct 0905won't runwon't run
Kimi K2 Thinkingwon't runwon't run
Kimi K2.5won't runwon't run
MiniMax M2won't runwon't run
MiniMax M2.1won't runwon't run
MiniMax M2.5won't runwon't run
MiniMax M3won't runwon't run
Mistral Medium 3.5 128Bwon't runwon't run

Questions

Which is better for local AI, the NVIDIA RTX PRO 4500 Blackwell or the NVIDIA Tesla P40?
Choose the NVIDIA RTX PRO 4500 Blackwell to run bigger models. Choose the NVIDIA RTX PRO 4500 Blackwell for speed.
Which can run bigger models, the NVIDIA RTX PRO 4500 Blackwell or the NVIDIA Tesla P40?
At 128k context the NVIDIA RTX PRO 4500 Blackwell runs 50 of the models we track fully in memory and the NVIDIA Tesla P40 31; only the NVIDIA RTX PRO 4500 Blackwell holds Qwen 3.8 27B.
Which is faster for local AI, the NVIDIA RTX PRO 4500 Blackwell or the NVIDIA Tesla P40?
On the models both run fully in memory, the NVIDIA RTX PRO 4500 Blackwell is about 2.6× as fast as the NVIDIA Tesla P40 (estimated, median across models).

More comparisons

More hardware for local AI