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 32k context it runs 57 of the 125 open models we track fully in memory; the most capable is Qwen 3.8 27B at Q5_K_M, at about 18 tok/s (estimated). We have no current price for the NVIDIA RTX PRO 4000 Blackwell in Germany.
Updated . Speeds are estimates; how they are worked out.
Prices
Every retailer we track in Germany, cheapest first. Buy links may earn us a commission.
No retailer in this market lists it right now.
What it runs
Every model in the catalogue, with the best quant for this hardware. Estimates, not measurements.
| Model | Quant | Fits | Memory | Max context | Decode @32k | Prefill | Good for |
|---|---|---|---|---|---|---|---|
| Qwen 3.8 27B Alibaba Qwen | Q5_K_M | fits | 22.5 GB | 44k78k with q8 cache | 18tok/sest. | 340 tok/s | chat |
| Qwen 3.6 27B Alibaba Qwen | Q5_K_M | fits | 21.2 GB | 64k112k with q8 cache | 19tok/sest. | 340 tok/s | chat |
| Gemma 4 31B | Q4_K_M | fits | 21.2 GB | 57k110k with q8 cache | 19tok/sest. | 303 tok/s | chat |
| Qwen 3.6 35B-A3B Alibaba Qwen | Q4_K_M | fits | 22.3 GB | 74k116k with q8 cache | 96tok/sest. | 2430 tok/s | chat, coding, agents, long documents |
| Qwen 3.5 27B Alibaba Qwen | Q5_K_M | fits | 21.3 GB | 62k110k with q8 cache | 19tok/sest. | 340 tok/s | chat |
| Gemma 4 26B-A4B | UD-Q6_K | fits | 23.3 GB | 36k62k with q8 cache | 56tok/sest. | 2162 tok/s | chat, coding, agents, long documents |
| Qwen 3.5 35B-A3B Alibaba Qwen | Q4_K_M | fits | 22.1 GB | 84k131k with q8 cache | 97tok/sest. | 2430 tok/s | chat, coding, agents, long documents |
| GLM 4.7 Flash Z.ai | Q5_K_M | fits | 22.5 GB | 48k82k with q8 cache | 47tok/sest. | 1269 tok/s | chat, coding, agents |
| Devstral Small 2 24B Mistral AI | Q5_K_M | fits | 21.5 GB | 43k79k with q8 cache | 19tok/sest. | 359 tok/s | chat |
| Nemotron 3.5 Lightning 30B-A3B NVIDIA | MXFP4 | fits | 22.7 GB | 88k115k with q8 cache | 102tok/sest. | 2773 tok/s | chat, coding, agents, long documents |
| gpt-oss 20B OpenAI | MXFP4 | fits | 12.9 GB | 128k | 77tok/sest. | 1814 tok/s | chat, coding, agents, long documents |
| Qwen3 Coder 30B-A3B Alibaba Qwen | Q4_K_M | fits | 21.2 GB | 54k97k with q8 cache | 50tok/sest. | 1118 tok/s | chat, coding, agents |
| Gemma 4 12B | Q8_0 | fits | 13.5 GB | 256k | 31tok/sest. | 789 tok/s | chat, coding, long documents |
| Qwen 3.5 9B Alibaba Qwen | Q8_0 | fits | 10.8 GB | 256k | 39tok/sest. | 984 tok/s | chat, coding, long documents |
| Mistral Small 3.2 24B Mistral AI | Q5_K_M | fits | 21.5 GB | 43k79k with q8 cache | 19tok/sest. | 359 tok/s | chat |
| Gemma 3 27B | Q5_K_M | fits | 21.7 GB | 51k95k with q8 cache | 19tok/sest. | 361 tok/s | chat |
| Ministral 3 14B Mistral AI | Q8_0 | fits | 19.3 GB | 57k104k with q8 cache | 21tok/sest. | 546 tok/s | chat, coding |
| Qwen3 14B Alibaba Qwen | Q8_0 | fits | 20.5 GB | 40k | 20tok/sest. | 491 tok/s | chat |
| DeepSeek R1 0528 Qwen3 8B DeepSeek | Q8_0 | fits | 13.5 GB | 99k128k with q8 cache | 31tok/sest. | 810 tok/s | chat, coding |
| Qwen3 8B Alibaba Qwen | Q8_0 | fits | 13.5 GB | 40k | 31tok/sest. | 810 tok/s | chat, coding |
| Phi-4 14B Microsoft | Q8_0 | fits | 21.6 GB | 16k | 19tok/sest. | 494 tok/s | chat |
| Qwen 3.5 4B Alibaba Qwen | Q8_0 | fits | 6.1 GB | 256k | 74tok/sest. | 1841 tok/s | chat, coding, long documents |
| Ministral 3 8B Mistral AI | Q8_0 | fits | 13.5 GB | 103k187k with q8 cache | 31tok/sest. | 783 tok/s | chat, coding |
| Granite 4.2 8B IBM | Q8_0 | fits | 14.6 GB | 86k128k with q8 cache | 28tok/sest. | 745 tok/s | chat, coding |
| Gemma 4 E4B | Q8_0 | fits | 9.4 GB | 128k | 45tok/sest. | 1236 tok/s | chat, coding, long documents |
| LFM2.5 8B-A1B Liquid AI | Q8_0 | fits | 10.8 GB | 125k | 94tok/sest. | 4042 tok/s | chat, coding, long documents |
| Llama 3.1 8B Meta | Q6_K | fits | 11.0 GB | 126k128k with q8 cache | 38tok/sest. | 856 tok/s | chat, coding, long documents |
| Qwen2.5 0.5B Alibaba Qwen | Q8_0 | fits | 1.8 GB | 32k | 446tok/sest. | 8804 tok/s | chat, coding, long documents |
| Qwen3.5 0.8B Alibaba Qwen | Q8_0 | fits | 2.0 GB | 256k | 338tok/sest. | 8269 tok/s | chat, coding, agents, long documents |
| Qwen3.5 2B Alibaba Qwen | Q5_K_M | fits | 2.6 GB | 256k | 224tok/sest. | 3942 tok/s | chat, coding, agents, long documents |
| DeepSeek R1 Distill Qwen 1.5B DeepSeek | Q8_0 | fits | 3.5 GB | 128k | 147tok/sest. | 3312 tok/s | chat, coding, long documents |
| MiniCPM5 2B openbmb | Q8_0 | fits | 4.7 GB | 128k | 102tok/sest. | 1978 tok/s | chat, coding, agents, long documents |
| Spark X2.5 4B XHToken | mlx-4bit | fits | 4.9 GB | 483k808k with q8 cache | 97tok/sest. | 1942 tok/s | chat, coding, agents, long documents |
| LFM2.5 2.6B Liquid AI | Q8_0 | fits | 5.4 GB | 128k | 85tok/sest. | 2240 tok/s | chat, coding, agents, long documents |
| LFM2.5 VL 3B Liquid AI | Q8_0 | fits | 5.4 GB | 32k | 85tok/sest. | 2054 tok/s | chat, coding, agents, long documents |
| Mellum2.1 12B A2.5B Thinking JetBrains | Q8_0 | fits | 13.4 GB | 128k | 79tok/sest. | 2919 tok/s | chat, coding, agents, long documents |
| FrogNano 4B 2609 Microsoft | Q8_0 | fits | 6.2 GB | 256k | 73tok/sest. | 1841 tok/s | chat, coding, agents, long documents |
| granite vision 4.1 4b IBM | Q8_0 | fits | 6.8 GB | 128k | 66tok/sest. | 1435 tok/s | chat, coding, long documents |
| granite 4.2 3b IBM | Q8_0 | fits | 7.0 GB | 128k | 63tok/sest. | 1504 tok/s | chat, coding, agents, long documents |
| Kimi VL A3B Thinking 2506 Moonshot AI | Q8_0 | fits | 17.6 GB | 128k | 61tok/sest. | 2193 tok/s | chat, coding, long documents |
| Kimi VL A3B Moonshot AI | mlx-8bit | fits | 18.4 GB | 128k | 58tok/sest. | 2193 tok/s | chat, coding, agents, long documents |
| Ministral 3 3B Instruct 2512 Mistral AI | Q8_0 | fits | 7.5 GB | 179k256k with q8 cache | 58tok/sest. | 1438 tok/s | chat, coding, agents, long documents |
| LFM2.5 24B A2B Liquid AI | Q6_K | fits | 23.1 GB | 35k63k with q8 cache | 56tok/sest. | 2253 tok/s | chat, coding, agents |
| GLM 4 9B 0414 Z.ai | mlx-5bit | fits | 9.8 GB | 32k | 43tok/sest. | 715 tok/s | chat, coding, agents, long documents |
| DeepSeek R1 Distill Qwen 7B DeepSeek | Q8_0 | fits | 10.2 GB | 128k | 42tok/sest. | 968 tok/s | chat, coding, long documents |
| GLM 4.6V Flash Z.ai | Q8_0 | fits | 11.5 GB | 128k | 37tok/sest. | 674 tok/s | chat, coding, agents, long documents |
| DeepSeek R1 Distill Llama 8B DeepSeek | Q8_0 | fits | 12.8 GB | 112k128k with q8 cache | 33tok/sest. | 856 tok/s | chat, coding |
| GLM 4.1V 9B Thinking Z.ai | mlx-8bit | fits | 13.1 GB | 64k | 32tok/sest. | 674 tok/s | chat, coding, long documents |
| Ministral 3 8B Reasoning 2512 Mistral AI | Q8_0 | fits | 13.5 GB | 103k187k with q8 cache | 31tok/sest. | 783 tok/s | chat, coding, agents |
| Ministral 3 14B Reasoning 2512 Mistral AI | Q8_0 | fits | 19.3 GB | 57k104k with q8 cache | 21tok/sest. | 546 tok/s | chat, coding |
| DeepSeek R1 Distill Qwen 14B DeepSeek | Q8_0 | fits | 21.5 GB | 41k76k with q8 cache | 19tok/sest. | 462 tok/s | chat |
| Devstral Small 2507 Mistral AI | Q5_K_M | fits | 21.5 GB | 43k79k with q8 cache | 19tok/sest. | 365 tok/s | chat |
| Magistral Small 2506 Mistral AI | Q5_K_M | fits | 21.5 GB | 40k | 19tok/sest. | 365 tok/s | chat |
| Magistral Small 2507 Mistral AI | Q5_K_M | fits | 21.5 GB | 40k | 19tok/sest. | 365 tok/s | chat |
| Magistral Small 2509 Mistral AI | Q5_K_M | fits | 21.5 GB | 43k79k with q8 cache | 19tok/sest. | 359 tok/s | chat |
| Qwen3.8 27B OBLITERATED OBLITERATUS | mlx-4bit | fits | 22.3 GB | 48k84k with q8 cache | 18tok/sest. | 340 tok/s | chat |
| Muse Glimmer 30B meta-models | Q6_K | fits | 23.2 GB | 44k67k with q8 cache | 17tok/sest. | 328 tok/s | chat |
| Qwen 3.5 122B-A10B Alibaba Qwen | UD-Q2_K_XL | offload | 40.7 GB | – | 15tok/sest. | 590 tok/s | chat |
| Qwen3 Coder Next Alibaba Qwen | Q4_K_M | offload | 46.9 GB | – | 17tok/sest. | 1334 tok/s | chat |
| Mistral Small 4 119B Mistral AI | UD-Q2_K_XL | offload | 39.0 GB | – | 21tok/sest. | 669 tok/s | chat, coding |
| GLM 4.5 Air Z.ai | Q2_K | offload | 48.8 GB | – | 5tok/sest. | 225 tok/s | – |
| Qwen3 Next 80B-A3B Alibaba Qwen | Q4_K_M | offload | 46.9 GB | – | 21tok/sest. | 1588 tok/s | chat, coding |
| Kimi Linear 48B-A3B Moonshot AI | Q4_K_M | offload | 29.2 GB | – | 49tok/sest. | 1838 tok/s | chat, coding, long documents |
| Qwen3 32B Alibaba Qwen | Q4_K_M | offload | 27.3 GB | 16k30k with q8 cache | 7tok/sest. | 193 tok/s | chat |
| DeepSeek R1 Distill Qwen 32B DeepSeek | Q4_K_M | offload | 27.4 GB | 16k30k with q8 cache | 7tok/sest. | 221 tok/s | chat |
| granite 4.2 30b IBM | Q4_K_M | offload | 25.7 GB | 23k42k with q8 cache | 10tok/sest. | 263 tok/s | chat |
| DeepSeek V4 Flash DeepSeek | UD-Q2_K_XL | won't run | 93.8 GB | – | – | – | – |
| GLM 5.3 Flash Z.ai | UD-Q2_K_XL | won't run | 102.6 GB | – | – | – | – |
| Qwen 3.8 Flash Next Alibaba Qwen | UD-Q2_K_XL | won't run | 75.2 GB | – | – | – | – |
| MiniMax M2.7 MiniMax | UD-Q2_K_XL | won't run | 78.8 GB | – | – | – | – |
| Nemotron 3 Super 120B-A12B NVIDIA | UD-Q2_K_XL | won't run | 52.2 GB | – | – | – | – |
| gpt-oss 120B OpenAI | Q4_K_M | won't run | 60.5 GB | – | – | – | – |
| Llama 3.3 70B Meta | Q4_K_M | won't run | 50.5 GB | – | – | – | – |
| DeepSeek R1 DeepSeek | Q4_K_M | won't run | 379.7 GB | – | – | – | – |
| DeepSeek R1 0528 DeepSeek | Q4_K_M | won't run | 380.2 GB | – | – | – | – |
| DeepSeek R1 Distill Llama 70B DeepSeek | Q4_K_M | won't run | 50.5 GB | – | – | – | – |
| DeepSeek R1 Zero DeepSeek | Q4_K_M | won't run | 379.7 GB | – | – | – | – |
| DeepSeek V3 DeepSeek | Q4_K_M | won't run | 379.7 GB | – | – | – | – |
| DeepSeek V3 0324 DeepSeek | Q4_K_M | won't run | 379.7 GB | – | – | – | – |
| DeepSeek V3.1 DeepSeek | Q4_K_M | won't run | 380.6 GB | – | – | – | – |
| DeepSeek V3.1 Terminus DeepSeek | Q4_K_M | won't run | 380.6 GB | – | – | – | – |
| DeepSeek V3.2 DeepSeek | Q4_K_M | won't run | 380.6 GB | – | – | – | – |
| DeepSeek V4 Flash 0731 DeepSeek | MXFP4 | won't run | 149.2 GB | – | – | – | – |
| DeepSeek V4 Flash Vision Exp DeepSeek | MXFP4 | won't run | 158.0 GB | – | – | – | – |
| Devstral 2 123B Instruct 2512 Mistral AI | Q4_K_M | won't run | 81.6 GB | – | – | – | – |
| GLM 4.5 Z.ai | Q4_K_M | won't run | 214.0 GB | – | – | – | – |
| GLM 4.5V Z.ai | Q4_K_M | won't run | 65.9 GB | – | – | – | – |
| GLM 4.6 Z.ai | Q4_K_M | won't run | 213.2 GB | – | – | – | – |
| GLM 4.6V Z.ai | Q4_K_M | won't run | 72.2 GB | – | – | – | – |
| GLM 4.7 Z.ai | Q4_K_M | won't run | 214.0 GB | – | – | – | – |
| GLM 5 Z.ai | MXFP4 | won't run | 386.0 GB | – | – | – | – |
| GLM 5.1 Z.ai | MXFP4 | won't run | 423.6 GB | – | – | – | – |
| GLM 5.2 Z.ai | UD-Q4_K_M | won't run | 437.5 GB | – | – | – | – |
| GLM 5.3 Z.ai | Q8_0 | won't run | 750.0 GB | – | – | – | – |
| Kimi Dev 72B Moonshot AI | Q6_K | won't run | 70.8 GB | – | – | – | – |
| Kimi K2 Moonshot AI | Q4_K_M | won't run | 581.2 GB | – | – | – | – |
| Kimi K2 Instruct 0905 Moonshot AI | Q4_K_M | won't run | 581.6 GB | – | – | – | – |
| Kimi K2 Thinking Moonshot AI | Q4_K_M | won't run | 581.6 GB | – | – | – | – |
| Kimi K2.5 Moonshot AI | Q4_K_M | won't run | 581.6 GB | – | – | – | – |
| MiniMax M2 MiniMax | Q4_K_M | won't run | 137.5 GB | – | – | – | – |
| MiniMax M2.1 MiniMax | Q4_K_M | won't run | 137.5 GB | – | – | – | – |
| MiniMax M2.5 MiniMax | MXFP4 | won't run | 123.9 GB | – | – | – | – |
| MiniMax M3 MiniMax | MXFP4 | won't run | 244.3 GB | – | – | – | – |
| Mistral Medium 3.5 128B Mistral AI | Q4_K_M | won't run | 81.6 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 (Q5_K_M) fits fully in its memory at 32k context and generates at about 18 tok/s, an estimate from its memory bandwidth. 57 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 94 GB at 32k 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 Germany?
- We have no current price for the NVIDIA RTX PRO 4000 Blackwell in Germany.
Speeds on this page are for 32k context. See the best hardware for local AI at every budget.