readygpu

NVIDIA GeForce RTX 5090 vs NVIDIA RTX PRO 5000 Blackwell 72GB

For running models locally, priced in the US. At 128k context the first runs 50 models fully in memory, the second 68.

In short

The NVIDIA RTX PRO 5000 Blackwell 72GB has more memory for models: 72 GB against 32 GB. At 128k context the NVIDIA RTX PRO 5000 Blackwell 72GB runs 68 of the models we track fully in memory and the NVIDIA GeForce RTX 5090 50; only the NVIDIA RTX PRO 5000 Blackwell 72GB holds Qwen 3.5 122B-A10B. On the models both run fully in memory, the NVIDIA GeForce RTX 5090 is about 1.3× as fast as the NVIDIA RTX PRO 5000 Blackwell 72GB (estimated, median across models).

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

NVIDIA GeForce RTX 5090

$4,921

eBay

Add to build
NVIDIA RTX PRO 5000 Blackwell 72GB

Not priced here

Add to build

Specifications

SpecNVIDIA GeForce RTX 5090NVIDIA RTX PRO 5000 Blackwell 72GB
GPUGeForce RTX 5090RTX PRO 5000 Blackwell 72GB
Memory32 GB GDDR772 GB GDDR7
Memory bus512-bit384-bit
Memory bandwidth1,792 GB/s1,344 GB/s
FP16 compute209 TFLOPS73.69 TFLOPS
SoftwareCUDACUDA
Board power575 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 GeForce RTX 5090NVIDIA RTX PRO 5000 Blackwell 72GB
Qwen 3.8 27B36tok/sUD-Q6_K22tok/sQ8_0
Qwen 3.5 122B-A10Boffload63tok/sQ3_K_M
Qwen 3.6 27B36tok/sQ6_K22tok/sQ8_0
Gemma 4 31B37tok/sQ4_K_M19tok/sQ8_0
Qwen3 Coder Nextoffload77tok/sQ6_K
Nemotron 3 Super 120B-A12Boffload69tok/sUD-Q3_K_M
gpt-oss 120Bwon't run63tok/sMXFP4
Mistral Small 4 119Boffload69tok/sMXFP4
Qwen 3.6 35B-A3B128tok/sUD-Q6_K86tok/sQ8_0
Qwen 3.5 27B36tok/sQ6_K22tok/sQ8_0
Gemma 4 26B-A4B95tok/sQ8_071tok/sQ8_0
GLM 4.5 Airwon't run16tok/sQ2_K
Qwen 3.5 35B-A3B129tok/sQ6_K86tok/sQ8_0
GLM 4.7 Flash60tok/sQ6_K40tok/sQ8_0
Qwen3 Next 80B-A3Boffload88tok/sQ6_K
Devstral Small 2 24Boffload18tok/sQ8_0
Nemotron 3.5 Lightning 30B-A3B179tok/sUD-Q5_K_M120tok/sUD-Q6_K_XL
Kimi Linear 48B-A3B227tok/sQ4_K_M117tok/sQ8_0
gpt-oss 20B118tok/sMXFP489tok/sMXFP4
Qwen3 Coder 30B-A3B45tok/sQ4_K_M31tok/sQ8_0
Qwen3 32Boffload12tok/sQ8_0
Gemma 4 12B73tok/sQ8_055tok/sQ8_0
Qwen 3.5 9B80tok/sQ8_060tok/sQ8_0
Gemma 3 27B37tok/sQ5_K_M21tok/sQ8_0
Mistral Small 3.2 24Boffload18tok/sQ8_0
DeepSeek R1 Distill Qwen 32Boffload12tok/sQ8_0
Ministral 3 14B36tok/sQ5_K_M23tok/sQ8_0
Qwen3 14B35tok/sQ5_K_M22tok/sQ8_0
DeepSeek R1 0528 Qwen3 8B40tok/sQ8_030tok/sQ8_0
Qwen3 8B40tok/sQ8_030tok/sQ8_0
Phi-4 14Boffload20tok/sQ8_0
Qwen 3.5 4B126tok/sQ8_094tok/sQ8_0
Ministral 3 8B41tok/sQ8_031tok/sQ8_0
Granite 4.2 8B36tok/sQ8_027tok/sQ8_0
Gemma 4 E4B93tok/sQ8_070tok/sQ8_0
LFM2.5 8B-A1B89tok/sQ8_067tok/sQ8_0
Llama 3.1 8B47tok/sQ6_K35tok/sQ6_K
Qwen2.5 0.5B519tok/sQ8_0389tok/sQ8_0
Qwen3.5 0.8B455tok/sQ8_0341tok/sQ8_0
Qwen3.5 2B363tok/sQ5_K_M272tok/sQ5_K_M
DeepSeek R1 Distill Qwen 1.5B197tok/sQ8_0147tok/sQ8_0
Mellum2.1 12B A2.5B Thinking146tok/sQ8_0109tok/sQ8_0
Spark X2.5 4B140tok/smlx-4bit105tok/smlx-4bit
MiniCPM5 2B134tok/sQ8_0100tok/sQ8_0
FrogNano 4B 2609124tok/sQ8_093tok/sQ8_0
LFM2.5 2.6B102tok/sQ8_076tok/sQ8_0
LFM2.5 VL 3B102tok/sQ8_076tok/sQ8_0
Kimi VL A3B Thinking 250693tok/sQ8_070tok/sQ8_0
Kimi VL A3B91tok/smlx-8bit68tok/smlx-8bit
GLM 4 9B 041482tok/smlx-5bit61tok/smlx-5bit
granite vision 4.1 4b77tok/sQ8_058tok/sQ8_0
granite 4.2 3b76tok/sQ8_057tok/sQ8_0
GLM 4.6V Flash72tok/sQ8_054tok/sQ8_0
DeepSeek R1 Distill Qwen 7B71tok/sQ8_053tok/sQ8_0
GLM 4.1V 9B Thinking65tok/smlx-8bit49tok/smlx-8bit
Ministral 3 3B Instruct 251263tok/sQ8_047tok/sQ8_0
LFM2.5 24B A2B53tok/sQ6_K39tok/sQ8_0
Muse Glimmer 30B44tok/sQ6_K24tok/sQ8_0
DeepSeek R1 Distill Llama 8B43tok/sQ8_032tok/sQ8_0
Ministral 3 8B Reasoning 251241tok/sQ8_031tok/sQ8_0
Qwen3.8 27B OBLITERATED38tok/smlx-4bit28tok/smlx-4bit
Ministral 3 14B Reasoning 251236tok/sQ5_K_M23tok/sQ8_0
DeepSeek R1 Distill Qwen 14Boffload20tok/sQ8_0
Devstral Small 2507offload18tok/sQ8_0
Magistral Small 2506offload18tok/sQ8_0
Magistral Small 2507offload18tok/sQ8_0
Magistral Small 2509offload18tok/sQ8_0
granite 4.2 30boffload13tok/sQ8_0
Qwen 3.8 Flash Nextwon't runoffload
Llama 3.3 70Bwon't runoffload
DeepSeek R1 Distill Llama 70Bwon't runoffload
GLM 4.5Vwon't runoffload
GLM 4.6Vwon't runoffload
DeepSeek V4 Flashwon't runwon't run
GLM 5.3 Flashwon't runwon't run
MiniMax M2.7won'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
Devstral 2 123B Instruct 2512won'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 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 GeForce RTX 5090 or the NVIDIA RTX PRO 5000 Blackwell 72GB?
Choose the NVIDIA RTX PRO 5000 Blackwell 72GB to run bigger models. Choose the NVIDIA GeForce RTX 5090 for speed.
Which can run bigger models, the NVIDIA GeForce RTX 5090 or the NVIDIA RTX PRO 5000 Blackwell 72GB?
At 128k context the NVIDIA RTX PRO 5000 Blackwell 72GB runs 68 of the models we track fully in memory and the NVIDIA GeForce RTX 5090 50; only the NVIDIA RTX PRO 5000 Blackwell 72GB holds Qwen 3.5 122B-A10B.
Which is faster for local AI, the NVIDIA GeForce RTX 5090 or the NVIDIA RTX PRO 5000 Blackwell 72GB?
On the models both run fully in memory, the NVIDIA GeForce RTX 5090 is about 1.3× as fast as the NVIDIA RTX PRO 5000 Blackwell 72GB (estimated, median across models).

More comparisons

More hardware for local AI