readygpu

NVIDIA RTX PRO 4500 Blackwell vs NVIDIA RTX PRO 5000 Blackwell 48GB

For running models locally, priced in the UK. At 8k context the first runs 61 models fully in memory, the second 68.

In short

The NVIDIA RTX PRO 5000 Blackwell 48GB has more memory for models: 48 GB against 32 GB. At 8k context the NVIDIA RTX PRO 5000 Blackwell 48GB runs 68 of the models we track fully in memory and the NVIDIA RTX PRO 4500 Blackwell 61; only the NVIDIA RTX PRO 5000 Blackwell 48GB holds Qwen 3.5 122B-A10B. On the models both run fully in memory, the NVIDIA RTX PRO 5000 Blackwell 48GB is about 1.5× as fast as the NVIDIA RTX PRO 4500 Blackwell (estimated, median across models). In the UK the NVIDIA RTX PRO 4500 Blackwell costs £4,587 at eBay and the NVIDIA RTX PRO 5000 Blackwell 48GB £5,087 at eBay: the NVIDIA RTX PRO 4500 Blackwell is £500 cheaper.

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

NVIDIA RTX PRO 4500 Blackwell

£4,587

eBay

Add to build
NVIDIA RTX PRO 5000 Blackwell 48GB

£5,087

eBay

Add to build

Specifications

SpecNVIDIA RTX PRO 4500 BlackwellNVIDIA RTX PRO 5000 Blackwell 48GB
GPURTX PRO 4500 BlackwellRTX PRO 5000 Blackwell 48GB
Memory32 GB GDDR748 GB GDDR7
Memory bus256-bit384-bit
Memory bandwidth896 GB/s1,344 GB/s
FP16 compute54.94 TFLOPS73.69 TFLOPS
SoftwareCUDACUDA
Board power200 W300 W
PCIe generationGen 5Gen 5
PCIe lanesx16x16

What each runs

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

ModelNVIDIA RTX PRO 4500 BlackwellNVIDIA RTX PRO 5000 Blackwell 48GB
Qwen 3.8 27B19tok/sQ8_028tok/sQ8_0
Qwen 3.5 122B-A10Boffload135tok/sUD-Q2_K_XL
Qwen 3.6 27B19tok/sQ8_029tok/sQ8_0
Gemma 4 31B21tok/sQ6_K24tok/sQ8_0
Qwen3 Coder Nextoffload190tok/sQ4_K_M
Mistral Small 4 119Boffload195tok/sUD-Q2_K_XL
Qwen 3.6 35B-A3B119tok/sQ6_K151tok/sQ8_0
Qwen 3.5 27B19tok/sQ8_029tok/sQ8_0
Gemma 4 26B-A4B73tok/sQ8_0110tok/sQ8_0
GLM 4.5 Airoffload69tok/sQ2_K
Qwen 3.5 35B-A3B126tok/sQ6_K151tok/sQ8_0
GLM 4.7 Flash58tok/sQ8_087tok/sQ8_0
Devstral Small 2 24B21tok/sQ8_032tok/sQ8_0
Qwen3 Next 80B-A3Boffload242tok/sQ4_K_M
Nemotron 3.5 Lightning 30B-A3B112tok/sUD-Q5_K_M146tok/sUD-Q6_K_XL
Kimi Linear 48B-A3B172tok/sQ4_K_M194tok/sQ6_K
gpt-oss 20B127tok/sMXFP4190tok/sMXFP4
Qwen3 Coder 30B-A3B102tok/sQ6_K125tok/sQ8_0
Qwen3 32B19tok/sQ6_K23tok/sQ8_0
Gemma 4 12B42tok/sQ8_063tok/sQ8_0
Qwen 3.5 9B56tok/sQ8_085tok/sQ8_0
Mistral Small 3.2 24B21tok/sQ8_032tok/sQ8_0
DeepSeek R1 Distill Qwen 32B19tok/sQ6_K23tok/sQ8_0
Gemma 3 27B19tok/sQ8_028tok/sQ8_0
Llama 3.3 70Boffload18tok/sQ4_K_M
Ministral 3 14B35tok/sQ8_053tok/sQ8_0
Qwen3 14B33tok/sQ8_049tok/sQ8_0
DeepSeek R1 0528 Qwen3 8B56tok/sQ8_084tok/sQ8_0
Qwen3 8B56tok/sQ8_084tok/sQ8_0
Phi-4 14B32tok/sQ8_048tok/sQ8_0
Qwen 3.5 4B116tok/sQ8_0174tok/sQ8_0
Ministral 3 8B55tok/sQ8_082tok/sQ8_0
Granite 4.2 8B52tok/sQ8_078tok/sQ8_0
Gemma 4 E4B66tok/sQ8_098tok/sQ8_0
LFM2.5 8B-A1B227tok/sQ8_0341tok/sQ8_0
Llama 3.1 8B72tok/sQ6_K109tok/sQ6_K
Qwen2.5 0.5B879tok/sQ8_01319tok/sQ8_0
Qwen3.5 0.8B596tok/sQ8_0895tok/sQ8_0
Qwen3.5 2B357tok/sQ5_K_M536tok/sQ5_K_M
DeepSeek R1 Distill Qwen 1.5B261tok/sQ8_0391tok/sQ8_0
MiniCPM5 2B183tok/sQ8_0275tok/sQ8_0
LFM2.5 2.6B164tok/sQ8_0247tok/sQ8_0
LFM2.5 VL 3B164tok/sQ8_0247tok/sQ8_0
Spark X2.5 4B163tok/smlx-4bit245tok/smlx-4bit
granite vision 4.1 4b130tok/sQ8_0194tok/sQ8_0
Ministral 3 3B Instruct 2512123tok/sQ8_0184tok/sQ8_0
granite 4.2 3b122tok/sQ8_0183tok/sQ8_0
LFM2.5 24B A2B119tok/sQ8_0179tok/sQ8_0
Mellum2.1 12B A2.5B Thinking118tok/sQ8_0177tok/sQ8_0
FrogNano 4B 2609112tok/sQ8_0169tok/sQ8_0
Kimi VL A3B Thinking 250699tok/sQ8_0149tok/sQ8_0
Kimi VL A3B95tok/smlx-8bit142tok/smlx-8bit
DeepSeek R1 Distill Qwen 7B65tok/sQ8_097tok/sQ8_0
GLM 4 9B 041465tok/smlx-5bit97tok/smlx-5bit
DeepSeek R1 Distill Llama 8B58tok/sQ8_087tok/sQ8_0
Ministral 3 8B Reasoning 251255tok/sQ8_082tok/sQ8_0
GLM 4.6V Flash54tok/sQ8_081tok/sQ8_0
GLM 4.1V 9B Thinking46tok/smlx-8bit69tok/smlx-8bit
Ministral 3 14B Reasoning 251235tok/sQ8_053tok/sQ8_0
DeepSeek R1 Distill Qwen 14B32tok/sQ8_048tok/sQ8_0
Qwen3.8 27B OBLITERATED26tok/smlx-4bit39tok/smlx-4bit
Devstral Small 250721tok/sQ8_032tok/sQ8_0
Magistral Small 250621tok/sQ8_032tok/sQ8_0
Magistral Small 250721tok/sQ8_032tok/sQ8_0
Magistral Small 250921tok/sQ8_032tok/sQ8_0
granite 4.2 30b21tok/sQ6_K25tok/sQ8_0
Muse Glimmer 30B17tok/sQ8_026tok/sQ8_0
DeepSeek R1 Distill Llama 70Boffload18tok/sQ4_K_M
MiniMax M2.7won't runoffload
Nemotron 3 Super 120B-A12Boffloadoffload
gpt-oss 120Bwon't runoffload
Devstral 2 123B Instruct 2512won't runoffload
GLM 4.5Vwon't runoffload
GLM 4.6Vwon't runoffload
Kimi Dev 72Bwon't runoffload
Mistral Medium 3.5 128Bwon't runoffload
DeepSeek V4 Flashwon't runwon't run
GLM 5.3 Flashwon't runwon't run
Qwen 3.8 Flash Nextwon't runwon't run
DeepSeek R1won't runwon't run
DeepSeek R1 0528won'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
GLM 4.5won't runwon't run
GLM 4.6won'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 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

Questions

Which is better for local AI, the NVIDIA RTX PRO 4500 Blackwell or the NVIDIA RTX PRO 5000 Blackwell 48GB?
Choose the NVIDIA RTX PRO 5000 Blackwell 48GB to run bigger models. Choose the NVIDIA RTX PRO 5000 Blackwell 48GB for speed. Choose the NVIDIA RTX PRO 4500 Blackwell to spend less.
Which can run bigger models, the NVIDIA RTX PRO 4500 Blackwell or the NVIDIA RTX PRO 5000 Blackwell 48GB?
At 8k context the NVIDIA RTX PRO 5000 Blackwell 48GB runs 68 of the models we track fully in memory and the NVIDIA RTX PRO 4500 Blackwell 61; only the NVIDIA RTX PRO 5000 Blackwell 48GB holds Qwen 3.5 122B-A10B.
Which is faster for local AI, the NVIDIA RTX PRO 4500 Blackwell or the NVIDIA RTX PRO 5000 Blackwell 48GB?
On the models both run fully in memory, the NVIDIA RTX PRO 5000 Blackwell 48GB is about 1.5× as fast as the NVIDIA RTX PRO 4500 Blackwell (estimated, median across models).
What is the price difference between the NVIDIA RTX PRO 4500 Blackwell and the NVIDIA RTX PRO 5000 Blackwell 48GB?
In the UK the NVIDIA RTX PRO 4500 Blackwell costs £4,587 at eBay and the NVIDIA RTX PRO 5000 Blackwell 48GB £5,087 at eBay: the NVIDIA RTX PRO 4500 Blackwell is £500 cheaper.

More comparisons

More hardware for local AI