readygpu

NVIDIA GeForce RTX 5090 vs NVIDIA RTX PRO 2000 Blackwell

For running models locally, priced in the US. At 64k context the first runs 59 models fully in memory, the second 32.

In short

The NVIDIA GeForce RTX 5090 has more memory for models: 32 GB against 16 GB. At 64k context the NVIDIA GeForce RTX 5090 runs 59 of the models we track fully in memory and the NVIDIA RTX PRO 2000 Blackwell 32; only the NVIDIA GeForce RTX 5090 holds Qwen 3.8 27B. On the models both run fully in memory, the NVIDIA GeForce RTX 5090 is about 6.2× as fast as the NVIDIA RTX PRO 2000 Blackwell (estimated, median across models). In the US the NVIDIA GeForce RTX 5090 costs $6,900 at eBay and the NVIDIA RTX PRO 2000 Blackwell $1,540 at eBay: the NVIDIA RTX PRO 2000 Blackwell is $5,360 cheaper.

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

NVIDIA GeForce RTX 5090

$6,900

eBay

Add to build
NVIDIA RTX PRO 2000 Blackwell

$1,540

eBay

Add to build

Specifications

SpecNVIDIA GeForce RTX 5090NVIDIA RTX PRO 2000 Blackwell
GPUGeForce RTX 5090RTX PRO 2000 Blackwell
Memory32 GB GDDR716 GB GDDR7
Memory bus512-bit128-bit
Memory bandwidth1,792 GB/s288 GB/s
FP16 compute209 TFLOPS17.03 TFLOPS
SoftwareCUDACUDA
Board power575 W70 W
PCIe generationGen 5Gen 5
PCIe lanesx16x8

What each runs

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

ModelNVIDIA GeForce RTX 5090NVIDIA RTX PRO 2000 Blackwell
Qwen 3.8 27B39tok/sQ6_Koffload
Qwen 3.6 27B41tok/sQ6_Koffload
Gemma 4 31B35tok/sQ6_Koffload
Qwen 3.6 35B-A3B167tok/sQ6_Koffload
Qwen 3.5 27B41tok/sQ6_Koffload
Gemma 4 26B-A4B117tok/sQ8_0offload
Qwen 3.5 35B-A3B175tok/sQ6_Koffload
GLM 4.7 Flash87tok/sQ6_Koffload
Devstral Small 2 24B37tok/sQ6_Koffload
Nemotron 3.5 Lightning 30B-A3B201tok/sUD-Q5_K_Moffload
Kimi Linear 48B-A3B277tok/sQ4_K_Moffload
gpt-oss 20B165tok/sMXFP427tok/sMXFP4
Qwen3 Coder 30B-A3B75tok/sQ6_Koffload
Gemma 4 12B79tok/sQ8_013tok/sQ8_0
Qwen 3.5 9B95tok/sQ8_015tok/sQ8_0
Gemma 3 27B40tok/sQ6_Koffload
Mistral Small 3.2 24B37tok/sQ6_Koffload
Ministral 3 14B44tok/sQ8_0offload
Qwen3 14B42tok/sQ8_0offload
DeepSeek R1 0528 Qwen3 8B61tok/sQ8_012tok/sQ4_K_M
Qwen3 8B61tok/sQ8_012tok/sQ4_K_M
Phi-4 14B38tok/sQ8_0offload
Qwen 3.5 4B166tok/sQ8_027tok/sQ8_0
Ministral 3 8B61tok/sQ8_012tok/sQ5_K_M
Granite 4.2 8B55tok/sQ8_0offload
Gemma 4 E4B110tok/sQ8_018tok/sQ8_0
LFM2.5 8B-A1B156tok/sQ8_025tok/sQ8_0
Llama 3.1 8B73tok/sQ6_K12tok/sQ6_K
Qwen2.5 0.5B831tok/sQ8_0134tok/sQ8_0
Qwen3.5 0.8B679tok/sQ8_0109tok/sQ8_0
Qwen3.5 2B492tok/sQ5_K_M79tok/sQ5_K_M
LFM3 1.7B418tok/sQ8_067tok/sQ8_0
DeepSeek R1 Distill Qwen 1.5B294tok/sQ8_047tok/sQ8_0
MiniCPM5 2B202tok/sQ8_033tok/sQ8_0
Spark X2.5 4B202tok/smlx-4bit32tok/smlx-4bit
Mellum2.1 12B A2.5B Thinking183tok/sQ8_029tok/sQ8_0
FrogNano 4B 2609163tok/sQ8_026tok/sQ8_0
LFM2.5 2.6B161tok/sQ8_026tok/sQ8_0
LFM2.5 VL 3B161tok/sQ8_026tok/sQ8_0
Kimi VL A3B Thinking 2506130tok/sQ8_027tok/sQ4_K_M
Kimi VL A3B126tok/smlx-8bit26tok/smlx-5bit
granite vision 4.1 4b124tok/sQ8_020tok/sQ8_0
granite 4.2 3b120tok/sQ8_019tok/sQ8_0
Ministral 3 3B Instruct 2512105tok/sQ8_017tok/sQ8_0
GLM 4 9B 0414102tok/smlx-5bit16tok/smlx-5bit
DeepSeek R1 Distill Qwen 7B94tok/sQ8_015tok/sQ8_0
LFM2.5 24B A2B89tok/sQ8_0offload
GLM 4.6V Flash88tok/sQ8_014tok/sQ8_0
GLM 4.1V 9B Thinking77tok/smlx-8bit12tok/smlx-8bit
DeepSeek R1 Distill Llama 8B65tok/sQ8_012tok/sQ6_K
Ministral 3 8B Reasoning 251261tok/sQ8_012tok/sQ5_K_M
Muse Glimmer 30B46tok/sQ6_Koffload
Ministral 3 14B Reasoning 251244tok/sQ8_0offload
Qwen3.8 27B OBLITERATED44tok/smlx-4bitoffload
Devstral Small 250740tok/sQ5_K_Moffload
Magistral Small 250740tok/sQ5_K_Moffload
Magistral Small 250940tok/sQ5_K_Moffload
DeepSeek R1 Distill Qwen 14B39tok/sQ8_0offload
Magistral Small 250637tok/sQ6_Koffload
Qwen 3.5 122B-A10Boffloadwon't run
Qwen3 Coder Nextoffloadwon't run
Nemotron 3 Super 120B-A12Boffloadwon't run
Mistral Small 4 119Boffloadoffload
GLM 4.5 Airoffloadwon't run
Qwen3 Next 80B-A3Boffloadwon't run
Qwen3 32Boffloadoffload
DeepSeek R1 Distill Qwen 32Boffloadoffload
granite 4.2 30boffloadoffload
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
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 GeForce RTX 5090 or the NVIDIA RTX PRO 2000 Blackwell?
Choose the NVIDIA GeForce RTX 5090 to run bigger models. Choose the NVIDIA GeForce RTX 5090 for speed. Choose the NVIDIA RTX PRO 2000 Blackwell to spend less.
Which can run bigger models, the NVIDIA GeForce RTX 5090 or the NVIDIA RTX PRO 2000 Blackwell?
At 64k context the NVIDIA GeForce RTX 5090 runs 59 of the models we track fully in memory and the NVIDIA RTX PRO 2000 Blackwell 32; only the NVIDIA GeForce RTX 5090 holds Qwen 3.8 27B.
Which is faster for local AI, the NVIDIA GeForce RTX 5090 or the NVIDIA RTX PRO 2000 Blackwell?
On the models both run fully in memory, the NVIDIA GeForce RTX 5090 is about 6.2× as fast as the NVIDIA RTX PRO 2000 Blackwell (estimated, median across models).
What is the price difference between the NVIDIA GeForce RTX 5090 and the NVIDIA RTX PRO 2000 Blackwell?
In the US the NVIDIA GeForce RTX 5090 costs $6,900 at eBay and the NVIDIA RTX PRO 2000 Blackwell $1,540 at eBay: the NVIDIA RTX PRO 2000 Blackwell is $5,360 cheaper.

More comparisons

More hardware for local AI