readygpu

NVIDIA GeForce RTX 4070 Ti Super vs AMD Radeon RX 9070

For running models locally, priced in the US. At 8k context the first runs 43 models fully in memory, the second 37.

In short

Both have 16 GB of memory for models; the NVIDIA GeForce RTX 4070 Ti Super has 672 GB/s of bandwidth and the AMD Radeon RX 9070 640 GB/s. At 8k context the NVIDIA GeForce RTX 4070 Ti Super runs 43 of the models we track fully in memory and the AMD Radeon RX 9070 37; only the NVIDIA GeForce RTX 4070 Ti Super holds Devstral Small 2 24B. On the models both run fully in memory, the NVIDIA GeForce RTX 4070 Ti Super is about 1.2× as fast as the AMD Radeon RX 9070 (estimated, median across models).

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

NVIDIA GeForce RTX 4070 Ti Super

$1,099

eBay

Add to build
AMD Radeon RX 9070

Not priced here

Add to build

Specifications

SpecNVIDIA GeForce RTX 4070 Ti SuperAMD Radeon RX 9070
GPUGeForce RTX 4070 Ti SuperRadeon RX 9070
Memory16 GB GDDR6X16 GB GDDR6
Memory bus256-bit256-bit
Memory bandwidth672 GB/s640 GB/s
FP16 compute88 TFLOPS36.1 TFLOPS
SoftwareCUDAROCm
Board power285 W220 W
PCIe generationGen 4Gen 5
PCIe lanesx16x16

What each runs

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

ModelNVIDIA GeForce RTX 4070 Ti SuperAMD Radeon RX 9070
Devstral Small 2 24B27tok/sQ4_K_Moffload
gpt-oss 20B95tok/sMXFP480tok/sMXFP4
Gemma 4 12B32tok/sQ8_027tok/sQ8_0
Qwen 3.5 9B42tok/sQ8_036tok/sQ8_0
Mistral Small 3.2 24B27tok/sQ4_K_Moffload
Qwen3 14B31tok/sQ6_K26tok/sQ6_K
Ministral 3 14B27tok/sQ8_028tok/sQ6_K
DeepSeek R1 0528 Qwen3 8B42tok/sQ8_036tok/sQ8_0
Qwen3 8B42tok/sQ8_036tok/sQ8_0
Phi-4 14B30tok/sQ6_K26tok/sQ6_K
Qwen 3.5 4B87tok/sQ8_073tok/sQ8_0
Ministral 3 8B41tok/sQ8_035tok/sQ8_0
Granite 4.2 8B39tok/sQ8_033tok/sQ8_0
Gemma 4 E4B49tok/sQ8_042tok/sQ8_0
LFM2.5 8B-A1B170tok/sQ8_0144tok/sQ8_0
Llama 3.1 8B54tok/sQ6_K46tok/sQ6_K
Qwen2.5 0.5B660tok/sQ8_0557tok/sQ8_0
Qwen3.5 0.8B447tok/sQ8_0378tok/sQ8_0
Qwen3.5 2B268tok/sQ5_K_M226tok/sQ5_K_M
DeepSeek R1 Distill Qwen 1.5B196tok/sQ8_0165tok/sQ8_0
MiniCPM5 2B137tok/sQ8_0116tok/sQ8_0
LFM2.5 2.6B123tok/sQ8_0104tok/sQ8_0
LFM2.5 VL 3B123tok/sQ8_0104tok/sQ8_0
Spark X2.5 4B123tok/smlx-4bit104tok/smlx-4bit
Kimi VL A3B Thinking 2506115tok/sQ4_K_M97tok/sQ4_K_M
granite vision 4.1 4b97tok/sQ8_082tok/sQ8_0
Ministral 3 3B Instruct 251292tok/sQ8_078tok/sQ8_0
granite 4.2 3b91tok/sQ8_077tok/sQ8_0
Kimi VL A3B90tok/smlx-6bit76tok/smlx-6bit
Mellum2.1 12B A2.5B Thinking89tok/sQ8_075tok/sQ8_0
FrogNano 4B 260984tok/sQ8_071tok/sQ8_0
DeepSeek R1 Distill Qwen 7B49tok/sQ8_041tok/sQ8_0
GLM 4 9B 041449tok/smlx-5bit41tok/smlx-5bit
DeepSeek R1 Distill Llama 8B43tok/sQ8_037tok/sQ8_0
Ministral 3 8B Reasoning 251241tok/sQ8_035tok/sQ8_0
GLM 4.6V Flash40tok/sQ8_034tok/sQ8_0
GLM 4.1V 9B Thinking34tok/smlx-8bit29tok/smlx-8bit
DeepSeek R1 Distill Qwen 14B30tok/sQ6_K26tok/sQ6_K
Devstral Small 250727tok/sQ4_K_Moffload
Magistral Small 250627tok/sQ4_K_Moffload
Magistral Small 250727tok/sQ4_K_Moffload
Magistral Small 250927tok/sQ4_K_Moffload
Ministral 3 14B Reasoning 251227tok/sQ8_028tok/sQ6_K
Qwen 3.8 27Boffloadoffload
Qwen 3.5 122B-A10Boffloadoffload
Qwen 3.6 27Boffloadoffload
Gemma 4 31Boffloadoffload
Mistral Small 4 119Boffloadoffload
Qwen 3.6 35B-A3Boffloadoffload
Qwen 3.5 27Boffloadoffload
Gemma 4 26B-A4Boffloadoffload
Qwen 3.5 35B-A3Boffloadoffload
GLM 4.7 Flashoffloadoffload
Nemotron 3.5 Lightning 30B-A3Boffloadoffload
Qwen3 Coder 30B-A3Boffloadoffload
Kimi Linear 48B-A3Boffloadoffload
Qwen3 32Boffloadoffload
Gemma 3 27Boffloadoffload
DeepSeek R1 Distill Qwen 32Boffloadoffload
LFM2.5 24B A2Boffloadoffload
Muse Glimmer 30Boffloadoffload
granite 4.2 30boffloadoffload
Qwen3.8 27B OBLITERATEDoffloadoffload
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
Nemotron 3 Super 120B-A12Bwon't runwon't run
Qwen3 Coder Nextwon't runwon't run
gpt-oss 120Bwon't runwon't run
GLM 4.5 Airwon't runwon't run
Qwen3 Next 80B-A3Bwon'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 NVIDIA GeForce RTX 4070 Ti Super or the AMD Radeon RX 9070?
Choose the NVIDIA GeForce RTX 4070 Ti Super to run bigger models. Choose the NVIDIA GeForce RTX 4070 Ti Super for speed.
Which can run bigger models, the NVIDIA GeForce RTX 4070 Ti Super or the AMD Radeon RX 9070?
At 8k context the NVIDIA GeForce RTX 4070 Ti Super runs 43 of the models we track fully in memory and the AMD Radeon RX 9070 37; only the NVIDIA GeForce RTX 4070 Ti Super holds Devstral Small 2 24B.
Which is faster for local AI, the NVIDIA GeForce RTX 4070 Ti Super or the AMD Radeon RX 9070?
On the models both run fully in memory, the NVIDIA GeForce RTX 4070 Ti Super is about 1.2× as fast as the AMD Radeon RX 9070 (estimated, median across models).

More comparisons

More hardware for local AI