readygpu

AMD Radeon PRO W7900 vs NVIDIA GeForce RTX 3090 Ti

For running models locally, priced in Canada. At 8k context the first runs 68 models fully in memory, the second 60.

In short

The AMD Radeon PRO W7900 has more memory for models: 48 GB against 24 GB. At 8k context the AMD Radeon PRO W7900 runs 68 of the models we track fully in memory and the NVIDIA GeForce RTX 3090 Ti 60; only the AMD Radeon PRO W7900 holds Qwen 3.5 122B-A10B. On the models both run fully in memory, the NVIDIA GeForce RTX 3090 Ti is about 1.3× as fast as the AMD Radeon PRO W7900 (estimated, median across models). In Canada the AMD Radeon PRO W7900 costs $5,500 at eBay (used) and the NVIDIA GeForce RTX 3090 Ti $3,999 at Amazon: the NVIDIA GeForce RTX 3090 Ti is $1,501 cheaper.

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

AMD Radeon PRO W7900

$5,500

eBay, used (asking)

Add to build
NVIDIA GeForce RTX 3090 Ti

$3,999

Amazon

Add to build

Specifications

SpecAMD Radeon PRO W7900NVIDIA GeForce RTX 3090 Ti
GPURadeon PRO W7900GeForce RTX 3090 Ti
Memory48 GB GDDR624 GB GDDR6X
Memory bus384-bit384-bit
Memory bandwidth864 GB/s1,008 GB/s
FP16 compute122.6 TFLOPS80 TFLOPS
SoftwareROCmCUDA
Board power295 W450 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 GeForce RTX 3090 Ti
Qwen 3.8 27B16tok/sQ8_028tok/sUD-Q6_K
Qwen 3.5 122B-A10B77tok/sUD-Q2_K_XLoffload
Qwen 3.6 27B16tok/sQ8_027tok/sQ6_K
Qwen3 Coder Next108tok/sQ4_K_Moffload
Gemma 4 31B14tok/sQ8_027tok/sQ5_K_M
Mistral Small 4 119B111tok/sUD-Q2_K_XLoffload
Qwen 3.6 35B-A3B86tok/sQ8_0179tok/sQ4_K_M
Qwen 3.5 27B16tok/sQ8_027tok/sQ6_K
Gemma 4 26B-A4B63tok/sQ8_095tok/sUD-Q6_K
GLM 4.5 Air39tok/sQ2_Koffload
Qwen3 Next 80B-A3B138tok/sQ4_K_Moffload
Qwen 3.5 35B-A3B86tok/sQ8_0181tok/sQ4_K_M
GLM 4.7 Flash50tok/sQ8_094tok/sQ5_K_M
Devstral Small 2 24B18tok/sQ8_030tok/sQ6_K
Nemotron 3.5 Lightning 30B-A3B83tok/sUD-Q6_K_XL163tok/sMXFP4
Kimi Linear 48B-A3B111tok/sQ6_Koffload
gpt-oss 20B108tok/sMXFP4142tok/sMXFP4
Qwen3 Coder 30B-A3B71tok/sQ8_0128tok/sQ5_K_M
Qwen3 32B13tok/sQ8_029tok/sQ4_K_M
Gemma 4 12B36tok/sQ8_048tok/sQ8_0
Qwen 3.5 9B48tok/sQ8_064tok/sQ8_0
Mistral Small 3.2 24B18tok/sQ8_030tok/sQ6_K
Gemma 3 27B16tok/sQ8_027tok/sQ6_K
DeepSeek R1 Distill Qwen 32B13tok/sQ8_028tok/sQ4_K_M
Llama 3.3 70B11tok/sQ4_K_Moffload
Ministral 3 14B30tok/sQ8_040tok/sQ8_0
Qwen3 14B28tok/sQ8_037tok/sQ8_0
DeepSeek R1 0528 Qwen3 8B48tok/sQ8_063tok/sQ8_0
Qwen3 8B48tok/sQ8_063tok/sQ8_0
Phi-4 14B28tok/sQ8_036tok/sQ8_0
Qwen 3.5 4B99tok/sQ8_0130tok/sQ8_0
Ministral 3 8B47tok/sQ8_062tok/sQ8_0
Granite 4.2 8B45tok/sQ8_059tok/sQ8_0
Gemma 4 E4B56tok/sQ8_074tok/sQ8_0
LFM2.5 8B-A1B194tok/sQ8_0256tok/sQ8_0
Llama 3.1 8B62tok/sQ6_K82tok/sQ6_K
Qwen2.5 0.5B752tok/sQ8_0989tok/sQ8_0
Qwen3.5 0.8B510tok/sQ8_0671tok/sQ8_0
Qwen3.5 2B306tok/sQ5_K_M402tok/sQ5_K_M
DeepSeek R1 Distill Qwen 1.5B223tok/sQ8_0294tok/sQ8_0
MiniCPM5 2B157tok/sQ8_0206tok/sQ8_0
LFM2.5 2.6B141tok/sQ8_0185tok/sQ8_0
LFM2.5 VL 3B141tok/sQ8_0185tok/sQ8_0
Spark X2.5 4B140tok/smlx-4bit184tok/smlx-4bit
granite vision 4.1 4b111tok/sQ8_0146tok/sQ8_0
Ministral 3 3B Instruct 2512105tok/sQ8_0138tok/sQ8_0
granite 4.2 3b104tok/sQ8_0137tok/sQ8_0
LFM2.5 24B A2B102tok/sQ8_0154tok/sQ6_K
Mellum2.1 12B A2.5B Thinking101tok/sQ8_0133tok/sQ8_0
FrogNano 4B 260996tok/sQ8_0127tok/sQ8_0
Kimi VL A3B Thinking 250685tok/sQ8_0112tok/sQ8_0
Kimi VL A3B81tok/smlx-8bit107tok/smlx-8bit
DeepSeek R1 Distill Qwen 7B56tok/sQ8_073tok/sQ8_0
GLM 4 9B 041455tok/smlx-5bit73tok/smlx-5bit
DeepSeek R1 Distill Llama 8B49tok/sQ8_065tok/sQ8_0
Ministral 3 8B Reasoning 251247tok/sQ8_062tok/sQ8_0
GLM 4.6V Flash46tok/sQ8_061tok/sQ8_0
GLM 4.1V 9B Thinking39tok/smlx-8bit52tok/smlx-8bit
Ministral 3 14B Reasoning 251230tok/sQ8_040tok/sQ8_0
DeepSeek R1 Distill Qwen 14B27tok/sQ8_036tok/sQ8_0
Qwen3.8 27B OBLITERATED22tok/smlx-4bit29tok/smlx-4bit
Devstral Small 250718tok/sQ8_035tok/sQ5_K_M
Magistral Small 250618tok/sQ8_030tok/sQ6_K
Magistral Small 250718tok/sQ8_035tok/sQ5_K_M
Magistral Small 250918tok/sQ8_035tok/sQ5_K_M
Muse Glimmer 30B15tok/sQ8_027tok/sQ6_K
granite 4.2 30b14tok/sQ8_027tok/sQ5_K_M
DeepSeek R1 Distill Llama 70B11tok/sQ4_K_Moffload
MiniMax M2.7offloadwon't run
gpt-oss 120Boffloadwon't run
Nemotron 3 Super 120B-A12Boffloadwon't run
GLM 4.5Voffloadwon't run
GLM 4.6Voffloadwon't run
Kimi Dev 72Boffloadwon't run
DeepSeek V4 Flashwon't runwon't run
GLM 5.3 Flashwon't runwon't run
Qwen 3.8 Flash Nextwon'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
Devstral 2 123B Instruct 2512won'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
Mistral Medium 3.5 128Bwon't runwon't run

Questions

Which is better for local AI, the AMD Radeon PRO W7900 or the NVIDIA GeForce RTX 3090 Ti?
Choose the AMD Radeon PRO W7900 to run bigger models. Choose the NVIDIA GeForce RTX 3090 Ti for speed. Choose the NVIDIA GeForce RTX 3090 Ti to spend less.
Which can run bigger models, the AMD Radeon PRO W7900 or the NVIDIA GeForce RTX 3090 Ti?
At 8k context the AMD Radeon PRO W7900 runs 68 of the models we track fully in memory and the NVIDIA GeForce RTX 3090 Ti 60; only the AMD Radeon PRO W7900 holds Qwen 3.5 122B-A10B.
Which is faster for local AI, the AMD Radeon PRO W7900 or the NVIDIA GeForce RTX 3090 Ti?
On the models both run fully in memory, the NVIDIA GeForce RTX 3090 Ti is about 1.3× as fast as the AMD Radeon PRO W7900 (estimated, median across models).
What is the price difference between the AMD Radeon PRO W7900 and the NVIDIA GeForce RTX 3090 Ti?
In Canada the AMD Radeon PRO W7900 costs $5,500 at eBay (used) and the NVIDIA GeForce RTX 3090 Ti $3,999 at Amazon: the NVIDIA GeForce RTX 3090 Ti is $1,501 cheaper.

More comparisons

More hardware for local AI