GPUs & Macs › AMD Radeon RX 9070
AMD Radeon RX 9070: local LLM speed
Llama 2 7B (Q4_0) generates 120 tokens/s on the AMD Radeon RX 9070 (Vulkan), and processes prompts at 3,164 tokens/s.
Measured results
| Model | Generate tok/s | Prompt tok/s | Quant | Backend | Test | Source |
|---|---|---|---|---|---|---|
| Llama 2 7B | 120 | 3,164 | Q4_0 | Vulkan FA off | tg128 | llama.cpp #10879 details |
| Llama 2 7B | 120 | 2,860 | Q4_0 | Vulkan FA on | tg128 | llama.cpp #10879 details |
| Llama 2 7B | 115 | 2,453 | Q4_0 | ROCm FA on | tg128 | llama.cpp #15021 details |
| Llama 2 7B | 114 | 2,382 | Q4_0 | ROCm FA off | tg128 | llama.cpp #15021 details |
What fits on the AMD Radeon RX 9070
Estimates for this chip are scaled from a measured run: Llama 2 7B Q4_0 at 120 tok/s (Vulkan, source). Fit assumes the whole model in GPU memory with ~8K context.
| Model | Best quant that fits | File | Expected tok/s |
|---|---|---|---|
| Qwen3.8-27B | UD-IQ3_XXS | 10.9 GB | 41.8 estimate |
| gemma-4-12B-it | Q8_0 | 12.7 GB | 36.1 estimate |
| Qwen3.5-9B | Q8_0 | 9.5 GB | 48.0 estimate |
| Qwen3.5-4B | Q8_0 | 4.5 GB | 102 estimate |
| Qwen3.6-35B-A3B | UD-IQ3_XXS | 13.2 GB | 202 estimate |
| gemma-4-26B-A4B-it | UD-IQ4_XS | 13.6 GB | 109 estimate |
| MiMo-V2.6-Distill-Qwen-9B | Q8_0 | 9.5 GB | 47.9 estimate |
| gemma-4-E4B-it | Q8_0 | 8.2 GB | 55.9 estimate |
| Muse-Glimmer-30B | UD-IQ3_XXS | 13.1 GB | 34.8 estimate |
| Qwen3-4B | Q8_0 | 4.3 GB | 107 estimate |
| gemma-4-31B-it | UD-IQ3_XXS | 11.8 GB | 38.7 estimate |
| gemma-4-12B-it-qat-q4_0-unquantized | UD-Q4_K_XL | 6.7 GB | 68.1 estimate |
| gemma-3-1b-it | Q8_0 | 1.1 GB | 381 estimate |
| gemma-4-E2B-it | Q8_0 | 5.0 GB | 92.1 estimate |
Similar hardware
AMD Radeon RX 9070 XT · AMD Radeon RX 7800 XT · AMD Radeon RX 7900 GRE · AMD Radeon RX 6900 XT · AMD Radeon RX 6800 · AMD Radeon VII · AMD Radeon RX 6800 XT · AMD Radeon RX 9060 XT