readygpu

NVIDIA RTX A6000 vs NVIDIA RTX PRO 5000 Blackwell 72GB

For running models locally, priced in Germany. At 64k context the first runs 63 models fully in memory, the second 70.

In short

The NVIDIA RTX PRO 5000 Blackwell 72GB has more memory for models: 72 GB against 48 GB. At 64k context the NVIDIA RTX PRO 5000 Blackwell 72GB runs 70 of the models we track fully in memory and the NVIDIA RTX A6000 63; only the NVIDIA RTX PRO 5000 Blackwell 72GB holds Qwen3 Coder Next. On the models both run fully in memory, the NVIDIA RTX PRO 5000 Blackwell 72GB is about 1.7× as fast as the NVIDIA RTX A6000 (estimated, median across models).

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

NVIDIA RTX A6000

Not priced here

Add to build
NVIDIA RTX PRO 5000 Blackwell 72GB

Not priced here

Add to build

Specifications

SpecNVIDIA RTX A6000NVIDIA RTX PRO 5000 Blackwell 72GB
GPURTX A6000RTX PRO 5000 Blackwell 72GB
Memory48 GB GDDR672 GB GDDR7
Memory bus384-bit384-bit
Memory bandwidth768 GB/s1,344 GB/s
FP16 compute38.71 TFLOPS73.69 TFLOPS
SoftwareCUDACUDA
Board power300 W300 W
PCIe generationGen 4Gen 5
PCIe lanesx16x16

What each runs

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

ModelNVIDIA RTX A6000NVIDIA RTX PRO 5000 Blackwell 72GB
Qwen 3.8 27B14tok/sQ8_025tok/sQ8_0
Qwen 3.5 122B-A10B56tok/sUD-Q2_K_XL80tok/sQ3_K_M
Qwen 3.6 27B14tok/sQ8_025tok/sQ8_0
Gemma 4 31B12tok/sQ8_022tok/sQ8_0
Qwen3 Coder Nextoffload103tok/sQ6_K
Nemotron 3 Super 120B-A12Boffload75tok/sUD-Q3_K_M
gpt-oss 120Boffload90tok/sMXFP4
Mistral Small 4 119B74tok/sUD-Q2_K_XL85tok/sUD-Q4_K_M
Qwen 3.6 35B-A3B64tok/sQ8_0112tok/sQ8_0
Qwen 3.5 27B14tok/sQ8_025tok/sQ8_0
Gemma 4 26B-A4B50tok/sQ8_088tok/sQ8_0
GLM 4.5 Airoffload26tok/sQ3_K_M
Qwen 3.5 35B-A3B64tok/sQ8_0112tok/sQ8_0
GLM 4.7 Flash32tok/sQ8_057tok/sQ8_0
Devstral Small 2 24B13tok/sQ8_023tok/sQ8_0
Qwen3 Next 80B-A3Boffload122tok/sQ6_K
Nemotron 3.5 Lightning 30B-A3B76tok/sUD-Q6_K_XL132tok/sUD-Q6_K_XL
Kimi Linear 48B-A3B94tok/sQ6_K133tok/sQ8_0
gpt-oss 20B71tok/sMXFP4124tok/sMXFP4
Qwen3 Coder 30B-A3B30tok/sQ8_052tok/sQ8_0
Qwen3 32B11tok/sQ6_K16tok/sQ8_0
Gemma 4 12B34tok/sQ8_059tok/sQ8_0
Qwen 3.5 9B41tok/sQ8_071tok/sQ8_0
Gemma 3 27B14tok/sQ8_024tok/sQ8_0
Mistral Small 3.2 24B13tok/sQ8_023tok/sQ8_0
DeepSeek R1 Distill Qwen 32B11tok/sQ6_K16tok/sQ8_0
Llama 3.3 70Boffload12tok/sQ5_K_M
Ministral 3 14B19tok/sQ8_033tok/sQ8_0
Qwen3 14B18tok/sQ8_032tok/sQ8_0
DeepSeek R1 0528 Qwen3 8B26tok/sQ8_045tok/sQ8_0
Qwen3 8B26tok/sQ8_045tok/sQ8_0
Phi-4 14B16tok/sQ8_029tok/sQ8_0
Qwen 3.5 4B71tok/sQ8_0125tok/sQ8_0
Ministral 3 8B26tok/sQ8_046tok/sQ8_0
Granite 4.2 8B24tok/sQ8_042tok/sQ8_0
Gemma 4 E4B47tok/sQ8_082tok/sQ8_0
LFM2.5 8B-A1B67tok/sQ8_0117tok/sQ8_0
Llama 3.1 8B31tok/sQ6_K55tok/sQ6_K
Qwen2.5 0.5B356tok/sQ8_0624tok/sQ8_0
Qwen3.5 0.8B291tok/sQ8_0509tok/sQ8_0
Qwen3.5 2B211tok/sQ5_K_M369tok/sQ5_K_M
DeepSeek R1 Distill Qwen 1.5B126tok/sQ8_0221tok/sQ8_0
MiniCPM5 2B87tok/sQ8_0152tok/sQ8_0
Spark X2.5 4B86tok/smlx-4bit151tok/smlx-4bit
Mellum2.1 12B A2.5B Thinking78tok/sQ8_0137tok/sQ8_0
FrogNano 4B 260970tok/sQ8_0122tok/sQ8_0
LFM2.5 2.6B69tok/sQ8_0121tok/sQ8_0
LFM2.5 VL 3B69tok/sQ8_0121tok/sQ8_0
Kimi VL A3B Thinking 250656tok/sQ8_097tok/sQ8_0
Kimi VL A3B54tok/smlx-8bit94tok/smlx-8bit
granite vision 4.1 4b53tok/sQ8_093tok/sQ8_0
granite 4.2 3b51tok/sQ8_090tok/sQ8_0
Ministral 3 3B Instruct 251245tok/sQ8_078tok/sQ8_0
GLM 4 9B 041444tok/smlx-5bit76tok/smlx-5bit
DeepSeek R1 Distill Qwen 7B40tok/sQ8_070tok/sQ8_0
LFM2.5 24B A2B38tok/sQ8_067tok/sQ8_0
GLM 4.6V Flash38tok/sQ8_066tok/sQ8_0
GLM 4.1V 9B Thinking33tok/smlx-8bit58tok/smlx-8bit
DeepSeek R1 Distill Llama 8B28tok/sQ8_049tok/sQ8_0
Ministral 3 8B Reasoning 251226tok/sQ8_046tok/sQ8_0
Ministral 3 14B Reasoning 251219tok/sQ8_033tok/sQ8_0
Qwen3.8 27B OBLITERATED19tok/smlx-4bit33tok/smlx-4bit
DeepSeek R1 Distill Qwen 14B17tok/sQ8_029tok/sQ8_0
Muse Glimmer 30B14tok/sQ8_025tok/sQ8_0
Devstral Small 250713tok/sQ8_023tok/sQ8_0
Magistral Small 250613tok/sQ8_023tok/sQ8_0
Magistral Small 250713tok/sQ8_023tok/sQ8_0
Magistral Small 250913tok/sQ8_023tok/sQ8_0
granite 4.2 30b10tok/sQ8_017tok/sQ8_0
DeepSeek R1 Distill Llama 70Boffload12tok/sQ5_K_M
DeepSeek V4 Flashwon't runoffload
Qwen 3.8 Flash Nextwon't runoffload
MiniMax M2.7won't runoffload
GLM 4.5Voffloadoffload
Devstral 2 123B Instruct 2512won't runoffload
GLM 4.6Vwon't runoffload
Kimi Dev 72Bwon't runoffload
Mistral Medium 3.5 128Bwon't runoffload
GLM 5.3 Flashwon'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 A6000 or the NVIDIA RTX PRO 5000 Blackwell 72GB?
Choose the NVIDIA RTX PRO 5000 Blackwell 72GB to run bigger models. Choose the NVIDIA RTX PRO 5000 Blackwell 72GB for speed.
Which can run bigger models, the NVIDIA RTX A6000 or the NVIDIA RTX PRO 5000 Blackwell 72GB?
At 64k context the NVIDIA RTX PRO 5000 Blackwell 72GB runs 70 of the models we track fully in memory and the NVIDIA RTX A6000 63; only the NVIDIA RTX PRO 5000 Blackwell 72GB holds Qwen3 Coder Next.
Which is faster for local AI, the NVIDIA RTX A6000 or the NVIDIA RTX PRO 5000 Blackwell 72GB?
On the models both run fully in memory, the NVIDIA RTX PRO 5000 Blackwell 72GB is about 1.7× as fast as the NVIDIA RTX A6000 (estimated, median across models).

More comparisons

More hardware for local AI