NVIDIA RTX PRO 4000 Blackwell
24 GB · 672 GB/s · 145 W
In short
The NVIDIA RTX PRO 4000 Blackwell has 24 GB of GDDR7 memory with 672 GB/s of bandwidth, which sets how fast it generates text. At 64k context it runs 63 of the 125 open models we track fully in memory; the most capable is Qwen 3.8 27B at Q8_0, at about 12 tok/s (estimated). We have no current price for the NVIDIA RTX PRO 4000 Blackwell in the US.
Updated . Speeds are estimates; how they are worked out.
Prices
Every retailer we track in the US, cheapest first. Buy links may earn us a commission.
No retailer in this market lists it right now.
What two of them run
Every model in the catalogue, with the best quant for this hardware. Estimates, not measurements.
| Model | Quant | Fits | Memory | Max context | Decode @64k | Prefill | Good for |
|---|---|---|---|---|---|---|---|
| Qwen 3.8 27B Alibaba Qwen | Q8_0 | fits | 32.7 GB | 256k | 12tok/sest. | 308 tok/s | – |
| Qwen 3.5 122B-A10B Alibaba Qwen | UD-Q2_K_XL | fits | 42.1 GB | 224k256k with q8 cache | 49tok/sest. | 642 tok/s | chat, coding, agents, long documents |
| Qwen 3.6 27B Alibaba Qwen | Q8_0 | fits | 32.2 GB | 256k | 13tok/sest. | 308 tok/s | – |
| Gemma 4 31B | Q8_0 | fits | 37.7 GB | 172k256k with q8 cache | 11tok/sest. | 282 tok/s | – |
| Mistral Small 4 119B Mistral AI | UD-Q2_K_XL | fits | 40.3 GB | 296k471k with q8 cache | 64tok/sest. | 631 tok/s | chat, coding, agents, long documents |
| Qwen 3.6 35B-A3B Alibaba Qwen | Q8_0 | fits | 37.2 GB | 256k | 56tok/sest. | 1856 tok/s | chat, coding, agents, long documents |
| Qwen 3.5 27B Alibaba Qwen | Q8_0 | fits | 32.2 GB | 256k | 13tok/sest. | 308 tok/s | – |
| Gemma 4 26B-A4B | Q8_0 | fits | 27.9 GB | 256k | 44tok/sest. | 1901 tok/s | chat, coding, agents, long documents |
| Qwen 3.5 35B-A3B Alibaba Qwen | Q8_0 | fits | 37.2 GB | 256k | 56tok/sest. | 1856 tok/s | chat, coding, agents, long documents |
| GLM 4.7 Flash Z.ai | Q8_0 | fits | 34.4 GB | 198k | 28tok/sest. | 921 tok/s | chat, coding, agents, long documents |
| Devstral Small 2 24B Mistral AI | Q8_0 | fits | 34.8 GB | 137k251k with q8 cache | 12tok/sest. | 304 tok/s | chat |
| Nemotron 3.5 Lightning 30B-A3B NVIDIA | UD-Q6_K_XL | fits | 34.5 GB | 256k | 66tok/sest. | 2289 tok/s | chat, coding, agents, long documents |
| Kimi Linear 48B-A3B Moonshot AI | Q6_K | fits | 39.7 GB | 588k802k with q8 cache | 82tok/sest. | 1491 tok/s | chat, coding, long documents |
| gpt-oss 20B OpenAI | MXFP4 | fits | 14.3 GB | 128k | 62tok/sest. | 1421 tok/s | chat, coding, agents, long documents |
| Qwen3 Coder 30B-A3B Alibaba Qwen | Q8_0 | fits | 37.7 GB | 155k256k with q8 cache | 26tok/sest. | 664 tok/s | chat, coding, agents, long documents |
| Qwen3 32B Alibaba Qwen | Q6_K | fits | 42.5 GB | 40k | 10tok/sest. | 157 tok/s | – |
| Gemma 4 12B | Q8_0 | fits | 14.6 GB | 256k | 30tok/sest. | 731 tok/s | chat, coding, long documents |
| Qwen 3.5 9B Alibaba Qwen | Q8_0 | fits | 12.4 GB | 256k | 36tok/sest. | 894 tok/s | chat, coding, long documents |
| Gemma 3 27B | Q8_0 | fits | 33.6 GB | 128k | 12tok/sest. | 345 tok/s | – |
| Mistral Small 3.2 24B Mistral AI | Q8_0 | fits | 34.8 GB | 128k | 12tok/sest. | 304 tok/s | chat |
| DeepSeek R1 Distill Qwen 32B DeepSeek | Q6_K | fits | 42.5 GB | 80k128k with q8 cache | 10tok/sest. | 195 tok/s | – |
| Ministral 3 14B Mistral AI | Q8_0 | fits | 24.8 GB | 199k256k with q8 cache | 17tok/sest. | 428 tok/s | chat, coding |
| Qwen3 14B Alibaba Qwen | Q8_0 | fits | 26.1 GB | 40k | 16tok/sest. | 374 tok/s | chat |
| DeepSeek R1 0528 Qwen3 8B DeepSeek | Q8_0 | fits | 18.6 GB | 128k | 23tok/sest. | 591 tok/s | chat, coding |
| Qwen3 8B Alibaba Qwen | Q8_0 | fits | 18.6 GB | 40k | 23tok/sest. | 591 tok/s | chat, coding |
| Phi-4 14B Microsoft | Q8_0 | fits | 28.5 GB | 16k | 14tok/sest. | 376 tok/s | chat |
| Qwen 3.5 4B Alibaba Qwen | Q8_0 | fits | 7.7 GB | 256k | 62tok/sest. | 1550 tok/s | chat, coding, long documents |
| Ministral 3 8B Mistral AI | Q8_0 | fits | 18.4 GB | 256k | 23tok/sest. | 585 tok/s | chat, coding, long documents |
| Granite 4.2 8B IBM | Q8_0 | fits | 20.2 GB | 128k | 21tok/sest. | 540 tok/s | chat, coding |
| Gemma 4 E4B | Q8_0 | fits | 10.9 GB | 128k | 41tok/sest. | 1171 tok/s | chat, coding, long documents |
| LFM2.5 8B-A1B Liquid AI | Q8_0 | fits | 12.9 GB | 125k | 58tok/sest. | 2500 tok/s | chat, coding, long documents |
| Llama 3.1 8B Meta | Q6_K | fits | 15.6 GB | 128k | 27tok/sest. | 635 tok/s | chat, coding, long documents |
| Qwen2.5 0.5B Alibaba Qwen | Q8_0 | fits | 2.7 GB | 32k | 312tok/sest. | 5544 tok/s | chat, coding, long documents |
| Qwen3.5 0.8B Alibaba Qwen | Q8_0 | fits | 3.0 GB | 256k | 255tok/sest. | 6286 tok/s | chat, coding, agents, long documents |
| Qwen3.5 2B Alibaba Qwen | Q5_K_M | fits | 3.6 GB | 256k | 184tok/sest. | 3427 tok/s | chat, coding, agents, long documents |
| DeepSeek R1 Distill Qwen 1.5B DeepSeek | Q8_0 | fits | 5.0 GB | 128k | 110tok/sest. | 2296 tok/s | chat, coding, long documents |
| MiniCPM5 2B openbmb | Q8_0 | fits | 6.6 GB | 128k | 76tok/sest. | 1294 tok/s | chat, coding, agents, long documents |
| Spark X2.5 4B XHToken | mlx-4bit | fits | 6.6 GB | 1024k | 76tok/sest. | 1589 tok/s | chat, coding, agents, long documents |
| Mellum2.1 12B A2.5B Thinking JetBrains | Q8_0 | fits | 14.4 GB | 128k | 69tok/sest. | 2317 tok/s | chat, coding, agents, long documents |
| FrogNano 4B 2609 Microsoft | Q8_0 | fits | 7.8 GB | 256k | 61tok/sest. | 1550 tok/s | chat, coding, agents, long documents |
| LFM2.5 2.6B Liquid AI | Q8_0 | fits | 7.9 GB | 128k | 60tok/sest. | 1569 tok/s | chat, coding, agents, long documents |
| LFM2.5 VL 3B Liquid AI | Q8_0 | fits | 7.9 GB | 32k | 60tok/sest. | 1476 tok/s | chat, coding, agents, long documents |
| Kimi VL A3B Thinking 2506 Moonshot AI | Q8_0 | fits | 19.2 GB | 128k | 49tok/sest. | 1593 tok/s | chat, coding, long documents |
| Kimi VL A3B Moonshot AI | mlx-8bit | fits | 20.0 GB | 128k | 47tok/sest. | 1593 tok/s | chat, coding, agents, long documents |
| granite vision 4.1 4b IBM | Q8_0 | fits | 9.8 GB | 128k | 46tok/sest. | 986 tok/s | chat, coding, long documents |
| granite 4.2 3b IBM | Q8_0 | fits | 10.1 GB | 128k | 45tok/sest. | 1018 tok/s | chat, coding, agents, long documents |
| Ministral 3 3B Instruct 2512 Mistral AI | Q8_0 | fits | 11.4 GB | 256k | 39tok/sest. | 975 tok/s | chat, coding, agents, long documents |
| GLM 4 9B 0414 Z.ai | mlx-5bit | fits | 11.7 GB | 32k | 38tok/sest. | 524 tok/s | chat, coding, agents, long documents |
| DeepSeek R1 Distill Qwen 7B DeepSeek | Q8_0 | fits | 12.5 GB | 128k | 35tok/sest. | 744 tok/s | chat, coding, long documents |
| LFM2.5 24B A2B Liquid AI | Q8_0 | fits | 30.2 GB | 128k | 33tok/sest. | 1432 tok/s | chat, coding, agents, long documents |
| GLM 4.6V Flash Z.ai | Q8_0 | fits | 13.3 GB | 128k | 33tok/sest. | 502 tok/s | chat, coding, agents, long documents |
| GLM 4.1V 9B Thinking Z.ai | mlx-8bit | fits | 14.9 GB | 64k | 29tok/sest. | 502 tok/s | chat, coding, long documents |
| DeepSeek R1 Distill Llama 8B DeepSeek | Q8_0 | fits | 17.4 GB | 128k | 24tok/sest. | 635 tok/s | chat, coding, long documents |
| Ministral 3 8B Reasoning 2512 Mistral AI | Q8_0 | fits | 18.4 GB | 256k | 23tok/sest. | 585 tok/s | chat, coding, agents, long documents |
| Ministral 3 14B Reasoning 2512 Mistral AI | Q8_0 | fits | 24.8 GB | 199k256k with q8 cache | 17tok/sest. | 428 tok/s | chat, coding |
| Qwen3.8 27B OBLITERATED OBLITERATUS | mlx-4bit | fits | 24.9 GB | 256k | 17tok/sest. | 308 tok/s | chat |
| DeepSeek R1 Distill Qwen 14B DeepSeek | Q8_0 | fits | 28.1 GB | 128k | 15tok/sest. | 342 tok/s | chat |
| Muse Glimmer 30B meta-models | Q8_0 | fits | 32.5 GB | 128k | 13tok/sest. | 311 tok/s | – |
| Devstral Small 2507 Mistral AI | Q8_0 | fits | 34.8 GB | 128k | 12tok/sest. | 308 tok/s | chat |
| Magistral Small 2506 Mistral AI | Q8_0 | fits | 34.8 GB | 40k | 12tok/sest. | 308 tok/s | chat |
| Magistral Small 2507 Mistral AI | Q8_0 | fits | 34.8 GB | 40k | 12tok/sest. | 308 tok/s | chat |
| Magistral Small 2509 Mistral AI | Q8_0 | fits | 34.8 GB | 128k | 12tok/sest. | 304 tok/s | chat |
| granite 4.2 30b IBM | Q8_0 | fits | 46.5 GB | 65k120k with q8 cache | 9tok/sest. | 227 tok/s | – |
| Qwen3 Coder Next Alibaba Qwen | Q4_K_M | offload | 48.2 GB | 14k23k with q8 cache | 51tok/sest. | 1460 tok/s | chat, coding, agents, long documents |
| Nemotron 3 Super 120B-A12B NVIDIA | UD-Q2_K_XL | offload | 53.0 GB | – | 23tok/sest. | 692 tok/s | chat, coding, long documents |
| gpt-oss 120B OpenAI | Q4_K_M | offload | 62.2 GB | – | 17tok/sest. | 844 tok/s | chat, coding |
| GLM 4.5 Air Z.ai | Q2_K | offload | 55.2 GB | 18k34k with q8 cache | 7tok/sest. | 189 tok/s | chat |
| Qwen3 Next 80B-A3B Alibaba Qwen | Q4_K_M | offload | 48.2 GB | 15k24k with q8 cache | 60tok/sest. | 1665 tok/s | chat, coding, agents, long documents |
| Llama 3.3 70B Meta | Q4_K_M | offload | 61.1 GB | 19k35k with q8 cache | 3tok/sest. | 80 tok/s | – |
| GLM 4.5V Z.ai | Q4_K_M | offload | 72.2 GB | – | 4tok/sest. | 161 tok/s | – |
| DeepSeek R1 Distill Llama 70B DeepSeek | Q4_K_M | offload | 61.1 GB | 19k35k with q8 cache | 3tok/sest. | 80 tok/s | – |
| DeepSeek V4 Flash DeepSeek | UD-Q2_K_XL | won't run | 97.0 GB | – | – | – | – |
| GLM 5.3 Flash Z.ai | UD-Q2_K_XL | won't run | 103.5 GB | – | – | – | – |
| Qwen 3.8 Flash Next Alibaba Qwen | UD-Q2_K_XL | won't run | 76.5 GB | – | – | – | – |
| MiniMax M2.7 MiniMax | UD-Q2_K_XL | won't run | 87.1 GB | – | – | – | – |
| DeepSeek R1 DeepSeek | Q4_K_M | won't run | 382.4 GB | – | – | – | – |
| DeepSeek R1 0528 DeepSeek | Q4_K_M | won't run | 382.9 GB | – | – | – | – |
| DeepSeek R1 Zero DeepSeek | Q4_K_M | won't run | 382.4 GB | – | – | – | – |
| DeepSeek V3 DeepSeek | Q4_K_M | won't run | 382.4 GB | – | – | – | – |
| DeepSeek V3 0324 DeepSeek | Q4_K_M | won't run | 382.4 GB | – | – | – | – |
| DeepSeek V3.1 DeepSeek | Q4_K_M | won't run | 383.3 GB | – | – | – | – |
| DeepSeek V3.1 Terminus DeepSeek | Q4_K_M | won't run | 383.3 GB | – | – | – | – |
| DeepSeek V3.2 DeepSeek | Q4_K_M | won't run | 383.3 GB | – | – | – | – |
| DeepSeek V4 Flash 0731 DeepSeek | MXFP4 | won't run | 152.5 GB | – | – | – | – |
| DeepSeek V4 Flash Vision Exp DeepSeek | MXFP4 | won't run | 161.3 GB | – | – | – | – |
| Devstral 2 123B Instruct 2512 Mistral AI | Q4_K_M | won't run | 93.2 GB | – | – | – | – |
| GLM 4.5 Z.ai | Q4_K_M | won't run | 226.1 GB | – | – | – | – |
| GLM 4.6 Z.ai | Q4_K_M | won't run | 225.3 GB | – | – | – | – |
| GLM 4.6V Z.ai | Q4_K_M | won't run | 78.6 GB | – | – | – | – |
| GLM 4.7 Z.ai | Q4_K_M | won't run | 226.1 GB | – | – | – | – |
| GLM 5 Z.ai | MXFP4 | won't run | 389.3 GB | – | – | – | – |
| GLM 5.1 Z.ai | MXFP4 | won't run | 426.9 GB | – | – | – | – |
| GLM 5.2 Z.ai | UD-Q4_K_M | won't run | 440.8 GB | – | – | – | – |
| GLM 5.3 Z.ai | Q8_0 | won't run | 753.3 GB | – | – | – | – |
| Kimi Dev 72B Moonshot AI | Q6_K | won't run | 81.4 GB | – | – | – | – |
| Kimi K2 Moonshot AI | Q4_K_M | won't run | 583.9 GB | – | – | – | – |
| Kimi K2 Instruct 0905 Moonshot AI | Q4_K_M | won't run | 584.3 GB | – | – | – | – |
| Kimi K2 Thinking Moonshot AI | Q4_K_M | won't run | 584.3 GB | – | – | – | – |
| Kimi K2.5 Moonshot AI | Q4_K_M | won't run | 584.3 GB | – | – | – | – |
| MiniMax M2 MiniMax | Q4_K_M | won't run | 145.8 GB | – | – | – | – |
| MiniMax M2.1 MiniMax | Q4_K_M | won't run | 145.8 GB | – | – | – | – |
| MiniMax M2.5 MiniMax | MXFP4 | won't run | 132.2 GB | – | – | – | – |
| MiniMax M3 MiniMax | MXFP4 | won't run | 248.7 GB | – | – | – | – |
| Mistral Medium 3.5 128B Mistral AI | Q4_K_M | won't run | 93.2 GB | – | – | – | – |
Specifications
From the manufacturer's published specifications; sources are linked where we have them.
- GPU
- RTX PRO 4000 Blackwell
- Memory
- 24 GB GDDR7
- Memory bus
- 192-bit
- Memory bandwidth
- 672 GB/s
- FP16 compute
- 46.9 TFLOPS
- Software
- CUDA
- Board power
- 145 W
- PCIe generation
- Gen 5
- PCIe lanes
- x16
Compare with
Questions
- How much memory does the NVIDIA RTX PRO 4000 Blackwell have for AI models?
- The NVIDIA RTX PRO 4000 Blackwell has 24 GB of GDDR7 memory with 672 GB/s of bandwidth, which sets how fast it generates text.
- What is the best model the NVIDIA RTX PRO 4000 Blackwell can run locally?
- Qwen 3.8 27B (Q8_0) fits fully in its memory at 64k context and generates at about 12 tok/s, an estimate from its memory bandwidth. 63 of the 125 models we track fit at that context.
- Can the NVIDIA RTX PRO 4000 Blackwell run DeepSeek V4 Flash?
- No: even at UD-Q2_K_XL, DeepSeek V4 Flash needs about 97 GB at 64k context, more than the NVIDIA RTX PRO 4000 Blackwell and system RAM can hold.
- How much does the NVIDIA RTX PRO 4000 Blackwell cost in the US?
- We have no current price for the NVIDIA RTX PRO 4000 Blackwell in the US.
Speeds on this page are for 64k context. See the best hardware for local AI at every budget.