readygpu

AMD Radeon AI PRO R9700 vs NVIDIA GeForce RTX 4080 Super

For running models locally, priced in Canada. At 64k context the first runs 54 models fully in memory, the second 29.

In short

The AMD Radeon AI PRO R9700 has more memory for models: 32 GB against 16 GB. At 64k context the AMD Radeon AI PRO R9700 runs 54 of the models we track fully in memory and the NVIDIA GeForce RTX 4080 Super 29; only the AMD Radeon AI PRO R9700 holds Qwen 3.8 27B. On the models both run fully in memory, the NVIDIA GeForce RTX 4080 Super is about 1.3× as fast as the AMD Radeon AI PRO R9700 (estimated, median across models). In Canada the AMD Radeon AI PRO R9700 costs $2,800 at eBay (used) and the NVIDIA GeForce RTX 4080 Super $1,599 at eBay (used): the NVIDIA GeForce RTX 4080 Super is $1,201 cheaper.

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

AMD Radeon AI PRO R9700

$2,800

eBay, used (asking)

Add to build
NVIDIA GeForce RTX 4080 Super

$1,599

eBay, used (asking)

Add to build

Specifications

SpecAMD Radeon AI PRO R9700NVIDIA GeForce RTX 4080 Super
GPURadeon AI PRO R9700GeForce RTX 4080 Super
Memory32 GB GDDR616 GB GDDR6X
Memory bus256-bit256-bit
Memory bandwidth640 GB/s736 GB/s
FP16 compute191 TFLOPS104 TFLOPS
SoftwareROCmCUDA
Board power300 W320 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 4080 Super
Qwen 3.8 27B12tok/sQ6_Koffload
Qwen 3.6 27B13tok/sQ6_Koffload
Gemma 4 31B11tok/sQ6_Koffload
Qwen 3.6 35B-A3B55tok/sUD-Q6_Koffload
Qwen 3.5 27B13tok/sQ6_Koffload
Gemma 4 26B-A4B37tok/sQ8_0offload
Qwen 3.5 35B-A3B55tok/sQ6_Koffload
GLM 4.7 Flash28tok/sQ6_Koffload
Devstral Small 2 24B12tok/sQ6_Koffload
Nemotron 3.5 Lightning 30B-A3B64tok/sUD-Q5_K_Moffload
Kimi Linear 48B-A3B88tok/sQ4_K_Moffload
gpt-oss 20B52tok/sMXFP468tok/sMXFP4
Qwen3 Coder 30B-A3B24tok/sQ6_Koffload
Gemma 4 12B25tok/sQ8_032tok/sQ8_0
Qwen 3.5 9B30tok/sQ8_039tok/sQ8_0
Gemma 3 27B13tok/sQ6_Koffload
Mistral Small 3.2 24B12tok/sQ6_Koffload
Ministral 3 14B14tok/sQ8_0offload
Qwen3 14B13tok/sQ8_0offload
DeepSeek R1 0528 Qwen3 8B19tok/sQ8_031tok/sQ4_K_M
Qwen3 8B19tok/sQ8_031tok/sQ4_K_M
Phi-4 14B12tok/sQ8_0offload
Qwen 3.5 4B53tok/sQ8_068tok/sQ8_0
Ministral 3 8B19tok/sQ8_030tok/sQ5_K_M
Granite 4.2 8B18tok/sQ8_0offload
Gemma 4 E4B35tok/sQ8_045tok/sQ8_0
LFM2.5 8B-A1B49tok/sQ8_064tok/sQ8_0
Llama 3.1 8B23tok/sQ6_K30tok/sQ6_K
Qwen2.5 0.5B263tok/sQ8_0342tok/sQ8_0
Qwen3.5 0.8B215tok/sQ8_0279tok/sQ8_0
Qwen3.5 2B156tok/sQ5_K_M202tok/sQ5_K_M
DeepSeek R1 Distill Qwen 1.5B93tok/sQ8_0121tok/sQ8_0
MiniCPM5 2B64tok/sQ8_083tok/sQ8_0
Spark X2.5 4B64tok/smlx-4bit83tok/smlx-4bit
Mellum2.1 12B A2.5B Thinking58tok/sQ8_075tok/sQ8_0
LFM2.5 2.6B51tok/sQ8_066tok/sQ8_0
LFM2.5 VL 3B51tok/sQ8_066tok/sQ8_0
Kimi VL A3B Thinking 250641tok/sQ8_069tok/sQ4_K_M
Kimi VL A3B40tok/smlx-8bit65tok/smlx-5bit
granite vision 4.1 4b39tok/sQ8_051tok/sQ8_0
granite 4.2 3b38tok/sQ8_049tok/sQ8_0
GLM 4 9B 041432tok/smlx-5bit42tok/smlx-5bit
DeepSeek R1 Distill Qwen 7B30tok/sQ8_039tok/sQ8_0
LFM2.5 24B A2B28tok/sQ8_0offload
GLM 4.6V Flash28tok/sQ8_036tok/sQ8_0
GLM 4.1V 9B Thinking24tok/smlx-8bit32tok/smlx-8bit
DeepSeek R1 Distill Llama 8B21tok/sQ8_030tok/sQ6_K
Ministral 3 8B Reasoning 251219tok/sQ8_030tok/sQ5_K_M
Qwen3.8 27B OBLITERATED14tok/smlx-4bitoffload
Devstral Small 250713tok/sQ5_K_Moffload
Magistral Small 250713tok/sQ5_K_Moffload
Magistral Small 250913tok/sQ5_K_Moffload
DeepSeek R1 Distill Qwen 14B12tok/sQ8_0offload
Magistral Small 250612tok/sQ6_Koffload
Qwen 3.5 122B-A10Boffloadwon't run
Qwen3 Coder Nextoffloadwon't run
Nemotron 3 Super 120B-A12Boffloadwon't run
Mistral Small 4 119Boffloadoffload
GLM 4.5 Airoffloadwon't run
Qwen3 Next 80B-A3Boffloadwon't run
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
gpt-oss 120Bwon'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 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 4.5Vwon'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 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 AI PRO R9700 or the NVIDIA GeForce RTX 4080 Super?
Choose the AMD Radeon AI PRO R9700 to run bigger models. Choose the NVIDIA GeForce RTX 4080 Super for speed. Choose the NVIDIA GeForce RTX 4080 Super to spend less.
Which can run bigger models, the AMD Radeon AI PRO R9700 or the NVIDIA GeForce RTX 4080 Super?
At 64k context the AMD Radeon AI PRO R9700 runs 54 of the models we track fully in memory and the NVIDIA GeForce RTX 4080 Super 29; only the AMD Radeon AI PRO R9700 holds Qwen 3.8 27B.
Which is faster for local AI, the AMD Radeon AI PRO R9700 or the NVIDIA GeForce RTX 4080 Super?
On the models both run fully in memory, the NVIDIA GeForce RTX 4080 Super is about 1.3× 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 4080 Super?
In Canada the AMD Radeon AI PRO R9700 costs $2,800 at eBay (used) and the NVIDIA GeForce RTX 4080 Super $1,599 at eBay (used): the NVIDIA GeForce RTX 4080 Super is $1,201 cheaper.

More comparisons