readygpu

AMD Radeon PRO W7900 vs NVIDIA RTX 6000 Ada Generation

For running models locally, priced in Germany. At 128k context the first runs 60 models fully in memory, the second 60.

In short

Both have 48 GB of memory for models; the AMD Radeon PRO W7900 has 864 GB/s of bandwidth and the NVIDIA RTX 6000 Ada Generation 960 GB/s. At 128k context both run the same 60 models we track fully in memory. On the models both run fully in memory, the NVIDIA RTX 6000 Ada Generation is about 1.3× as fast as the AMD Radeon PRO W7900 (estimated, median across models).

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

AMD Radeon PRO W7900

Not priced here

Add to build
NVIDIA RTX 6000 Ada Generation

Not priced here

Add to build

Specifications

SpecAMD Radeon PRO W7900NVIDIA RTX 6000 Ada Generation
GPURadeon PRO W7900RTX 6000 Ada Generation
Memory48 GB GDDR648 GB GDDR6
Memory bus384-bit384-bit
Memory bandwidth864 GB/s960 GB/s
FP16 compute122.6 TFLOPS91.1 TFLOPS
SoftwareROCmCUDA
Board power295 W300 W
PCIe generationGen 4Gen 4
PCIe lanesx16x16

What each runs

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

ModelAMD Radeon PRO W7900NVIDIA RTX 6000 Ada Generation
Qwen 3.8 27B13tok/sQ8_016tok/sQ8_0
Qwen 3.5 122B-A10B42tok/sUD-Q2_K_XL53tok/sUD-Q2_K_XL
Qwen 3.6 27B13tok/sQ8_016tok/sQ8_0
Gemma 4 31B11tok/sQ8_014tok/sQ8_0
Mistral Small 4 119B53tok/sUD-Q2_K_XL66tok/sUD-Q2_K_XL
Qwen 3.6 35B-A3B49tok/sQ8_061tok/sQ8_0
Qwen 3.5 27B13tok/sQ8_016tok/sQ8_0
Gemma 4 26B-A4B40tok/sQ8_051tok/sQ8_0
Qwen 3.5 35B-A3B49tok/sQ8_061tok/sQ8_0
GLM 4.7 Flash23tok/sQ8_029tok/sQ8_0
Devstral Small 2 24B10tok/sQ8_013tok/sQ8_0
Nemotron 3.5 Lightning 30B-A3B68tok/sUD-Q6_K_XL86tok/sUD-Q6_K_XL
Kimi Linear 48B-A3B80tok/sQ6_K100tok/sQ6_K
gpt-oss 20B50tok/sMXFP463tok/sMXFP4
Qwen3 Coder 30B-A3B18tok/sQ8_022tok/sQ8_0
Gemma 4 12B31tok/sQ8_039tok/sQ8_0
Qwen 3.5 9B34tok/sQ8_043tok/sQ8_0
Gemma 3 27B12tok/sQ8_015tok/sQ8_0
Mistral Small 3.2 24B10tok/sQ8_013tok/sQ8_0
Ministral 3 14B13tok/sQ8_017tok/sQ8_0
Qwen3 14B13tok/sQ8_016tok/sQ8_0
DeepSeek R1 0528 Qwen3 8B17tok/sQ8_021tok/sQ8_0
Qwen3 8B17tok/sQ8_021tok/sQ8_0
Phi-4 14B11tok/sQ8_014tok/sQ8_0
Qwen 3.5 4B54tok/sQ8_067tok/sQ8_0
Ministral 3 8B17tok/sQ8_022tok/sQ8_0
Granite 4.2 8B15tok/sQ8_019tok/sQ8_0
Gemma 4 E4B40tok/sQ8_050tok/sQ8_0
LFM2.5 8B-A1B38tok/sQ8_048tok/sQ8_0
Llama 3.1 8B20tok/sQ6_K25tok/sQ6_K
Qwen2.5 0.5B222tok/sQ8_0278tok/sQ8_0
Qwen3.5 0.8B195tok/sQ8_0244tok/sQ8_0
Qwen3.5 2B155tok/sQ5_K_M194tok/sQ5_K_M
DeepSeek R1 Distill Qwen 1.5B84tok/sQ8_0105tok/sQ8_0
Mellum2.1 12B A2.5B Thinking62tok/sQ8_078tok/sQ8_0
Spark X2.5 4B60tok/smlx-4bit75tok/smlx-4bit
MiniCPM5 2B57tok/sQ8_072tok/sQ8_0
FrogNano 4B 260953tok/sQ8_066tok/sQ8_0
LFM2.5 2.6B44tok/sQ8_055tok/sQ8_0
LFM2.5 VL 3B44tok/sQ8_055tok/sQ8_0
Kimi VL A3B Thinking 250640tok/sQ8_050tok/sQ8_0
Kimi VL A3B39tok/smlx-8bit49tok/smlx-8bit
GLM 4 9B 041435tok/smlx-5bit44tok/smlx-5bit
granite vision 4.1 4b33tok/sQ8_042tok/sQ8_0
granite 4.2 3b33tok/sQ8_041tok/sQ8_0
GLM 4.6V Flash31tok/sQ8_039tok/sQ8_0
DeepSeek R1 Distill Qwen 7B30tok/sQ8_038tok/sQ8_0
GLM 4.1V 9B Thinking28tok/smlx-8bit35tok/smlx-8bit
Ministral 3 3B Instruct 251227tok/sQ8_034tok/sQ8_0
LFM2.5 24B A2B22tok/sQ8_028tok/sQ8_0
DeepSeek R1 Distill Llama 8B19tok/sQ8_023tok/sQ8_0
Ministral 3 8B Reasoning 251217tok/sQ8_022tok/sQ8_0
Qwen3.8 27B OBLITERATED16tok/smlx-4bit20tok/smlx-4bit
Muse Glimmer 30B14tok/sQ8_017tok/sQ8_0
Ministral 3 14B Reasoning 251213tok/sQ8_017tok/sQ8_0
DeepSeek R1 Distill Qwen 14B12tok/sQ8_014tok/sQ8_0
Devstral Small 250710tok/sQ8_013tok/sQ8_0
Magistral Small 250610tok/sQ8_013tok/sQ8_0
Magistral Small 250710tok/sQ8_013tok/sQ8_0
Magistral Small 250910tok/sQ8_013tok/sQ8_0
Qwen3 Coder Nextoffloadoffload
Nemotron 3 Super 120B-A12Boffloadoffload
gpt-oss 120Boffloadoffload
GLM 4.5 Airoffloadoffload
Qwen3 Next 80B-A3Boffloadoffload
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
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 AMD Radeon PRO W7900 or the NVIDIA RTX 6000 Ada Generation?
Choose the NVIDIA RTX 6000 Ada Generation for speed.
Which can run bigger models, the AMD Radeon PRO W7900 or the NVIDIA RTX 6000 Ada Generation?
At 128k context both run the same 60 models we track fully in memory.
Which is faster for local AI, the AMD Radeon PRO W7900 or the NVIDIA RTX 6000 Ada Generation?
On the models both run fully in memory, the NVIDIA RTX 6000 Ada Generation is about 1.3× as fast as the AMD Radeon PRO W7900 (estimated, median across models).

More comparisons

More hardware for local AI