AMD Radeon PRO W7900
48 GB · 864 GB/s · 295 W
In short
The AMD Radeon PRO W7900 has 48 GB of GDDR6 memory with 864 GB/s of bandwidth, which sets how fast it generates text. At 128k context it runs 60 of the 125 open models we track fully in memory; the most capable is Qwen 3.8 27B at Q8_0, at about 13 tok/s (estimated). We have no current price for the AMD Radeon PRO W7900 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 it runs
Every model in the catalogue, with the best quant for this hardware. Estimates, not measurements.
| Model | Quant | Fits | Memory | Max context | Decode @128k | Prefill | Good for |
|---|---|---|---|---|---|---|---|
| Qwen 3.8 27B Alibaba Qwen | Q8_0 | fits | 36.7 GB | 256k | 13tok/sest. | 528 tok/s | chat |
| Qwen 3.5 122B-A10B Alibaba Qwen | UD-Q2_K_XL | fits | 43.7 GB | 248k256k with q8 cache | 42tok/sest. | 938 tok/s | chat, coding, agents, long documents |
| Qwen 3.6 27B Alibaba Qwen | Q8_0 | fits | 36.3 GB | 256k | 13tok/sest. | 528 tok/s | chat |
| Gemma 4 31B | Q8_0 | fits | 42.7 GB | 181k256k with q8 cache | 11tok/sest. | 501 tok/s | – |
| Mistral Small 4 119B Mistral AI | UD-Q2_K_XL | fits | 41.8 GB | 321k511k with q8 cache | 53tok/sest. | 813 tok/s | chat, coding, agents, long documents |
| Qwen 3.6 35B-A3B Alibaba Qwen | Q8_0 | fits | 38.5 GB | 256k | 49tok/sest. | 2564 tok/s | chat, coding, agents, long documents |
| Qwen 3.5 27B Alibaba Qwen | Q8_0 | fits | 36.3 GB | 256k | 13tok/sest. | 528 tok/s | chat |
| Gemma 4 26B-A4B | Q8_0 | fits | 29.3 GB | 256k | 40tok/sest. | 3112 tok/s | chat, coding, agents, long documents |
| Qwen 3.5 35B-A3B Alibaba Qwen | Q8_0 | fits | 38.5 GB | 256k | 49tok/sest. | 2564 tok/s | chat, coding, agents, long documents |
| GLM 4.7 Flash Z.ai | Q8_0 | fits | 37.8 GB | 198k | 23tok/sest. | 1209 tok/s | chat, coding, agents, long documents |
| Devstral Small 2 24B Mistral AI | Q8_0 | fits | 44.9 GB | 142k259k with q8 cache | 10tok/sest. | 472 tok/s | chat |
| Nemotron 3.5 Lightning 30B-A3B NVIDIA | UD-Q6_K_XL | fits | 34.9 GB | 256k | 68tok/sest. | 3449 tok/s | chat, coding, agents, long documents |
| Kimi Linear 48B-A3B Moonshot AI | Q6_K | fits | 40.3 GB | 640k872k with q8 cache | 80tok/sest. | 1924 tok/s | chat, coding, long documents |
| gpt-oss 20B OpenAI | MXFP4 | fits | 15.8 GB | 128k | 50tok/sest. | 2015 tok/s | chat, coding, agents, long documents |
| Qwen3 Coder 30B-A3B Alibaba Qwen | Q8_0 | fits | 43.8 GB | 162k256k with q8 cache | 18tok/sest. | 746 tok/s | chat, coding, agents, long documents |
| Gemma 4 12B | Q8_0 | fits | 15.7 GB | 256k | 31tok/sest. | 1293 tok/s | chat, coding, long documents |
| Qwen 3.5 9B Alibaba Qwen | Q8_0 | fits | 14.5 GB | 256k | 34tok/sest. | 1538 tok/s | chat, coding, long documents |
| Gemma 3 27B | Q8_0 | fits | 38.7 GB | 128k | 12tok/sest. | 646 tok/s | chat |
| Mistral Small 3.2 24B Mistral AI | Q8_0 | fits | 44.9 GB | 128k | 10tok/sest. | 472 tok/s | chat |
| Ministral 3 14B Mistral AI | Q8_0 | fits | 34.9 GB | 203k256k with q8 cache | 13tok/sest. | 606 tok/s | chat, coding |
| Qwen3 14B Alibaba Qwen | Q8_0 | fits | 36.2 GB | 40k | 13tok/sest. | 516 tok/s | chat, coding |
| DeepSeek R1 0528 Qwen3 8B DeepSeek | Q8_0 | fits | 27.7 GB | 128k | 17tok/sest. | 780 tok/s | chat, coding, long documents |
| Qwen3 8B Alibaba Qwen | Q8_0 | fits | 27.7 GB | 40k | 17tok/sest. | 780 tok/s | chat, coding, long documents |
| Phi-4 14B Microsoft | Q8_0 | fits | 41.1 GB | 16k | 11tok/sest. | 517 tok/s | chat, coding |
| Qwen 3.5 4B Alibaba Qwen | Q8_0 | fits | 9.8 GB | 256k | 54tok/sest. | 2396 tok/s | chat, coding, long documents |
| Ministral 3 8B Mistral AI | Q8_0 | fits | 27.0 GB | 256k | 17tok/sest. | 790 tok/s | chat, coding, long documents |
| Granite 4.2 8B IBM | Q8_0 | fits | 30.3 GB | 128k | 15tok/sest. | 709 tok/s | chat, coding, agents, long documents |
| Gemma 4 E4B | Q8_0 | fits | 12.7 GB | 128k | 40tok/sest. | 2157 tok/s | chat, coding, long documents |
| LFM2.5 8B-A1B Liquid AI | Q8_0 | fits | 15.9 GB | 125k | 38tok/sest. | 2883 tok/s | chat, coding, long documents |
| Llama 3.1 8B Meta | Q6_K | fits | 23.7 GB | 128k | 20tok/sest. | 851 tok/s | chat, coding, long documents |
| Qwen2.5 0.5B Alibaba Qwen | Q8_0 | fits | 3.5 GB | 32k | 222tok/sest. | 6477 tok/s | chat, coding, long documents |
| Qwen3.5 0.8B Alibaba Qwen | Q8_0 | fits | 3.8 GB | 256k | 195tok/sest. | 8637 tok/s | chat, coding, agents, long documents |
| Qwen3.5 2B Alibaba Qwen | Q5_K_M | fits | 4.4 GB | 256k | 155tok/sest. | 5523 tok/s | chat, coding, agents, long documents |
| DeepSeek R1 Distill Qwen 1.5B DeepSeek | Q8_0 | fits | 6.8 GB | 128k | 84tok/sest. | 2894 tok/s | chat, coding, long documents |
| Mellum2.1 12B A2.5B Thinking JetBrains | Q8_0 | fits | 15.4 GB | 128k | 62tok/sest. | 3335 tok/s | chat, coding, agents, long documents |
| Spark X2.5 4B XHToken | mlx-4bit | fits | 8.9 GB | 1024k | 60tok/sest. | 2369 tok/s | chat, coding, agents, long documents |
| MiniCPM5 2B openbmb | Q8_0 | fits | 9.3 GB | 128k | 57tok/sest. | 1556 tok/s | chat, coding, agents, long documents |
| FrogNano 4B 2609 Microsoft | Q8_0 | fits | 9.9 GB | 256k | 53tok/sest. | 2396 tok/s | chat, coding, agents, long documents |
| LFM2.5 2.6B Liquid AI | Q8_0 | fits | 11.7 GB | 128k | 44tok/sest. | 1996 tok/s | chat, coding, agents, long documents |
| LFM2.5 VL 3B Liquid AI | Q8_0 | fits | 11.7 GB | 32k | 44tok/sest. | 1920 tok/s | chat, coding, agents, long documents |
| Kimi VL A3B Thinking 2506 Moonshot AI | Q8_0 | fits | 21.2 GB | 128k | 40tok/sest. | 2094 tok/s | chat, coding, long documents |
| Kimi VL A3B Moonshot AI | mlx-8bit | fits | 21.9 GB | 128k | 39tok/sest. | 2094 tok/s | chat, coding, agents, long documents |
| GLM 4 9B 0414 Z.ai | mlx-5bit | fits | 14.2 GB | 32k | 35tok/sest. | 695 tok/s | chat, coding, agents, long documents |
| granite vision 4.1 4b IBM | Q8_0 | fits | 14.9 GB | 128k | 33tok/sest. | 1232 tok/s | chat, coding, long documents |
| granite 4.2 3b IBM | Q8_0 | fits | 15.2 GB | 128k | 33tok/sest. | 1256 tok/s | chat, coding, agents, long documents |
| GLM 4.6V Flash Z.ai | Q8_0 | fits | 15.9 GB | 128k | 31tok/sest. | 675 tok/s | chat, coding, agents, long documents |
| DeepSeek R1 Distill Qwen 7B DeepSeek | Q8_0 | fits | 16.1 GB | 128k | 30tok/sest. | 1033 tok/s | chat, coding, long documents |
| GLM 4.1V 9B Thinking Z.ai | mlx-8bit | fits | 17.5 GB | 64k | 28tok/sest. | 675 tok/s | chat, coding, long documents |
| Ministral 3 3B Instruct 2512 Mistral AI | Q8_0 | fits | 18.0 GB | 256k | 27tok/sest. | 1205 tok/s | chat, coding, agents, long documents |
| LFM2.5 24B A2B Liquid AI | Q8_0 | fits | 35.3 GB | 128k | 22tok/sest. | 1684 tok/s | chat, coding, agents, long documents |
| DeepSeek R1 Distill Llama 8B DeepSeek | Q8_0 | fits | 25.5 GB | 128k | 19tok/sest. | 851 tok/s | chat, coding, long documents |
| Ministral 3 8B Reasoning 2512 Mistral AI | Q8_0 | fits | 27.0 GB | 256k | 17tok/sest. | 790 tok/s | chat, coding, agents, long documents |
| Qwen3.8 27B OBLITERATED OBLITERATUS | mlx-4bit | fits | 28.9 GB | 256k | 16tok/sest. | 528 tok/s | chat, coding, long documents |
| Muse Glimmer 30B meta-models | Q8_0 | fits | 33.4 GB | 128k | 14tok/sest. | 574 tok/s | – |
| Ministral 3 14B Reasoning 2512 Mistral AI | Q8_0 | fits | 34.9 GB | 203k256k with q8 cache | 13tok/sest. | 606 tok/s | chat, coding |
| DeepSeek R1 Distill Qwen 14B DeepSeek | Q8_0 | fits | 40.2 GB | 128k | 12tok/sest. | 457 tok/s | chat, coding |
| Devstral Small 2507 Mistral AI | Q8_0 | fits | 44.9 GB | 128k | 10tok/sest. | 476 tok/s | chat |
| Magistral Small 2506 Mistral AI | Q8_0 | fits | 44.9 GB | 40k | 10tok/sest. | 476 tok/s | chat |
| Magistral Small 2507 Mistral AI | Q8_0 | fits | 44.9 GB | 40k | 10tok/sest. | 476 tok/s | chat |
| Magistral Small 2509 Mistral AI | Q8_0 | fits | 44.9 GB | 128k | 10tok/sest. | 472 tok/s | chat |
| Qwen3 Coder Next Alibaba Qwen | Q4_K_M | offload | 49.8 GB | 38k61k with q8 cache | 35tok/sest. | 2012 tok/s | chat, coding, agents, long documents |
| Nemotron 3 Super 120B-A12B NVIDIA | UD-Q2_K_XL | offload | 53.6 GB | – | 22tok/sest. | 1219 tok/s | chat, coding, long documents |
| gpt-oss 120B OpenAI | Q4_K_M | offload | 64.5 GB | – | 11tok/sest. | 1161 tok/s | chat, coding |
| GLM 4.5 Air Z.ai | Q2_K | offload | 66.8 GB | 22k41k with q8 cache | 3tok/sest. | 201 tok/s | chat |
| Qwen3 Next 80B-A3B Alibaba Qwen | Q4_K_M | offload | 49.8 GB | 39k63k with q8 cache | 39tok/sest. | 2199 tok/s | chat, coding, agents, long documents |
| Qwen3 32B Alibaba Qwen | Q4_K_M | offload | 52.0 GB | 40k | 5tok/sest. | 200 tok/s | chat |
| DeepSeek R1 Distill Qwen 32B DeepSeek | Q4_K_M | offload | 52.0 GB | 109k128k with q8 cache | 5tok/sest. | 267 tok/s | chat |
| granite 4.2 30b IBM | Q4_K_M | offload | 50.3 GB | 115k128k with q8 cache | 6tok/sest. | 324 tok/s | chat |
| DeepSeek V4 Flash DeepSeek | UD-Q2_K_XL | won't run | 102.5 GB | – | – | – | – |
| GLM 5.3 Flash Z.ai | UD-Q2_K_XL | won't run | 104.3 GB | – | – | – | – |
| Qwen 3.8 Flash Next Alibaba Qwen | UD-Q2_K_XL | won't run | 78.1 GB | – | – | – | – |
| MiniMax M2.7 MiniMax | UD-Q2_K_XL | won't run | 102.7 GB | – | – | – | – |
| Llama 3.3 70B Meta | Q4_K_M | won't run | 81.2 GB | 21k39k with q8 cache | – | – | – |
| DeepSeek R1 DeepSeek | Q4_K_M | won't run | 386.8 GB | – | – | – | – |
| DeepSeek R1 0528 DeepSeek | Q4_K_M | won't run | 387.3 GB | – | – | – | – |
| DeepSeek R1 Distill Llama 70B DeepSeek | Q4_K_M | won't run | 81.2 GB | 21k39k with q8 cache | – | – | – |
| DeepSeek R1 Zero DeepSeek | Q4_K_M | won't run | 386.8 GB | – | – | – | – |
| DeepSeek V3 DeepSeek | Q4_K_M | won't run | 386.8 GB | – | – | – | – |
| DeepSeek V3 0324 DeepSeek | Q4_K_M | won't run | 386.8 GB | – | – | – | – |
| DeepSeek V3.1 DeepSeek | Q4_K_M | won't run | 387.7 GB | – | – | – | – |
| DeepSeek V3.1 Terminus DeepSeek | Q4_K_M | won't run | 387.7 GB | – | – | – | – |
| DeepSeek V3.2 DeepSeek | Q4_K_M | won't run | 387.7 GB | – | – | – | – |
| DeepSeek V4 Flash 0731 DeepSeek | MXFP4 | won't run | 157.9 GB | – | – | – | – |
| DeepSeek V4 Flash Vision Exp DeepSeek | MXFP4 | won't run | 166.7 GB | – | – | – | – |
| Devstral 2 123B Instruct 2512 Mistral AI | Q4_K_M | won't run | 115.3 GB | – | – | – | – |
| GLM 4.5 Z.ai | Q4_K_M | won't run | 249.1 GB | – | – | – | – |
| GLM 4.5V Z.ai | Q4_K_M | won't run | 83.8 GB | – | – | – | – |
| GLM 4.6 Z.ai | Q4_K_M | won't run | 248.3 GB | – | – | – | – |
| GLM 4.6V Z.ai | Q4_K_M | won't run | 90.2 GB | – | – | – | – |
| GLM 4.7 Z.ai | Q4_K_M | won't run | 249.1 GB | – | – | – | – |
| GLM 5 Z.ai | MXFP4 | won't run | 394.9 GB | – | – | – | – |
| GLM 5.1 Z.ai | MXFP4 | won't run | 432.5 GB | – | – | – | – |
| GLM 5.2 Z.ai | UD-Q4_K_M | won't run | 446.4 GB | – | – | – | – |
| GLM 5.3 Z.ai | Q8_0 | won't run | 758.8 GB | – | – | – | – |
| Kimi Dev 72B Moonshot AI | Q6_K | won't run | 101.5 GB | – | – | – | – |
| Kimi K2 Moonshot AI | Q4_K_M | won't run | 588.3 GB | – | – | – | – |
| Kimi K2 Instruct 0905 Moonshot AI | Q4_K_M | won't run | 588.7 GB | – | – | – | – |
| Kimi K2 Thinking Moonshot AI | Q4_K_M | won't run | 588.7 GB | – | – | – | – |
| Kimi K2.5 Moonshot AI | Q4_K_M | won't run | 588.7 GB | – | – | – | – |
| MiniMax M2 MiniMax | Q4_K_M | won't run | 161.4 GB | – | – | – | – |
| MiniMax M2.1 MiniMax | Q4_K_M | won't run | 161.4 GB | – | – | – | – |
| MiniMax M2.5 MiniMax | MXFP4 | won't run | 147.8 GB | – | – | – | – |
| MiniMax M3 MiniMax | MXFP4 | won't run | 256.2 GB | – | – | – | – |
| Mistral Medium 3.5 128B Mistral AI | Q4_K_M | won't run | 115.3 GB | – | – | – | – |
Specifications
From the manufacturer's published specifications; sources are linked where we have them.
- GPU
- Radeon PRO W7900
- Memory
- 48 GB GDDR6
- Memory bus
- 384-bit
- Memory bandwidth
- 864 GB/s
- FP16 compute
- 122.6 TFLOPS
- Software
- ROCm
- Board power
- 295 W
- PCIe generation
- Gen 4
- PCIe lanes
- x16
Compare with
Questions
- How much memory does the AMD Radeon PRO W7900 have for AI models?
- The AMD Radeon PRO W7900 has 48 GB of GDDR6 memory with 864 GB/s of bandwidth, which sets how fast it generates text.
- What is the best model the AMD Radeon PRO W7900 can run locally?
- Qwen 3.8 27B (Q8_0) fits fully in its memory at 128k context and generates at about 13 tok/s, an estimate from its memory bandwidth. 60 of the 125 models we track fit at that context.
- Can the AMD Radeon PRO W7900 run DeepSeek V4 Flash?
- No: even at UD-Q2_K_XL, DeepSeek V4 Flash needs about 102 GB at 128k context, more than the AMD Radeon PRO W7900 and system RAM can hold.
- How much does the AMD Radeon PRO W7900 cost in the US?
- We have no current price for the AMD Radeon PRO W7900 in the US.
Speeds on this page are for 128k context. See the best hardware for local AI at every budget.