readygpu

AMD Radeon AI PRO R9700 vs NVIDIA GeForce RTX 3090 Ti

For running models locally, priced in Germany. At 32k context the first runs 29 models fully in memory, the second 26.

In short

The AMD Radeon AI PRO R9700 has more memory for models: 32 GB against 24 GB. At 32k context the AMD Radeon AI PRO R9700 runs 29 of the models we track fully in memory and the NVIDIA GeForce RTX 3090 Ti 26; only the AMD Radeon AI PRO R9700 holds Kimi Linear 48B-A3B. On the models both run fully in memory, the NVIDIA GeForce RTX 3090 Ti is about 1.8× as fast as the AMD Radeon AI PRO R9700 (estimated, median across models). In Germany the AMD Radeon AI PRO R9700 costs 1.867 € at Amazon and the NVIDIA GeForce RTX 3090 Ti 2.500 € at Amazon: the AMD Radeon AI PRO R9700 is 633 € cheaper.

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

AMD Radeon AI PRO R9700

1.867 €

Amazon

Add to build
NVIDIA GeForce RTX 3090 Ti

2.500 €

Amazon

Add to build

Specifications

SpecAMD Radeon AI PRO R9700NVIDIA GeForce RTX 3090 Ti
GPURadeon AI PRO R9700GeForce RTX 3090 Ti
Memory32 GB GDDR624 GB GDDR6X
Memory bus256-bit384-bit
Memory bandwidth640 GB/s1,008 GB/s
FP16 compute191 TFLOPS80 TFLOPS
SoftwareROCmCUDA
Board power300 W450 W
PCIe generationGen 5Gen 4
PCIe lanesx16x16

What each runs

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

ModelAMD Radeon AI PRO R9700NVIDIA GeForce RTX 3090 Ti
Qwen 3.8 27B11tok/sQ8_028tok/sUD-Q5_K_M
Qwen 3.6 27B11tok/sQ8_029tok/sQ5_K_M
Gemma 4 31B12tok/sQ6_K29tok/sQ4_K_M
Qwen 3.6 35B-A3B66tok/sUD-Q6_K145tok/sUD-Q4_K_M
Qwen 3.5 27B11tok/sQ8_029tok/sQ5_K_M
Gemma 4 26B-A4B42tok/sQ8_084tok/sUD-Q6_K
Qwen 3.5 35B-A3B67tok/sQ6_K146tok/sQ4_K_M
GLM 4.7 Flash36tok/sQ6_K70tok/sQ5_K_M
Devstral Small 2 24B12tok/sQ8_028tok/sQ5_K_M
Kimi Linear 48B-A3B99tok/sQ4_K_Moffload
gpt-oss 20B65tok/sMXFP4116tok/sMXFP4
Qwen3 Coder 30B-A3B37tok/sQ6_K74tok/sQ4_K_M
Qwen3 32B11tok/sQ5_K_Moffload
Gemma 4 12B26tok/sQ8_046tok/sQ8_0
Qwen 3.5 9B33tok/sQ8_059tok/sQ8_0
Mistral Small 3.2 24B12tok/sQ8_028tok/sQ5_K_M
DeepSeek R1 Distill Qwen 32B11tok/sQ5_K_Moffload
Gemma 3 27B11tok/sQ8_028tok/sQ5_K_M
Ministral 3 14B18tok/sQ8_032tok/sQ8_0
Qwen3 14B17tok/sQ8_030tok/sQ8_0
DeepSeek R1 0528 Qwen3 8B26tok/sQ8_046tok/sQ8_0
Qwen3 8B26tok/sQ8_046tok/sQ8_0
Phi-4 14B16tok/sQ8_028tok/sQ8_0
Qwen 3.5 4B63tok/sQ8_0112tok/sQ8_0
Ministral 3 8B26tok/sQ8_046tok/sQ8_0
Granite 4.2 8B24tok/sQ8_043tok/sQ8_0
Gemma 4 E4B38tok/sQ8_068tok/sQ8_0
LFM2.5 8B-A1B79tok/sQ8_0140tok/sQ8_0
Llama 3.1 8B32tok/sQ6_K57tok/sQ6_K
Qwen 3.5 122B-A10Boffloadoffload
Qwen3 Coder Nextoffloadoffload
Nemotron 3 Super 120B-A12Boffloadwon't run
Mistral Small 4 119Boffloadoffload
GLM 4.5 Airoffloadoffload
Qwen3 Next 80B-A3Boffloadoffload
Nemotron 3.5 Lightning 30B-A3Boffloadoffload
Llama 3.3 70Boffloadwon'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
MiniMax M2.7won't runwon't run
gpt-oss 120Bwon't runwon't run

Questions

Which is better for local AI, the AMD Radeon AI PRO R9700 or the NVIDIA GeForce RTX 3090 Ti?
Choose the AMD Radeon AI PRO R9700 to run bigger models. Choose the NVIDIA GeForce RTX 3090 Ti for speed. Choose the AMD Radeon AI PRO R9700 to spend less.
Which can run bigger models, the AMD Radeon AI PRO R9700 or the NVIDIA GeForce RTX 3090 Ti?
At 32k context the AMD Radeon AI PRO R9700 runs 29 of the models we track fully in memory and the NVIDIA GeForce RTX 3090 Ti 26; only the AMD Radeon AI PRO R9700 holds Kimi Linear 48B-A3B.
Which is faster for local AI, the AMD Radeon AI PRO R9700 or the NVIDIA GeForce RTX 3090 Ti?
On the models both run fully in memory, the NVIDIA GeForce RTX 3090 Ti is about 1.8× as fast as the AMD Radeon AI PRO R9700 (estimated, median across models).
What is the price difference between the AMD Radeon AI PRO R9700 and the NVIDIA GeForce RTX 3090 Ti?
In Germany the AMD Radeon AI PRO R9700 costs 1.867 € at Amazon and the NVIDIA GeForce RTX 3090 Ti 2.500 € at Amazon: the AMD Radeon AI PRO R9700 is 633 € cheaper.

More comparisons