GPUs & Macs › AMD Radeon RX 6800 XT

AMD Radeon RX 6800 XT: local LLM speed

Llama 2 7B (Q4_0) generates 100 tokens/s on the AMD Radeon RX 6800 XT (Vulkan), and processes prompts at 1,705 tokens/s.

Measured results

ModelGenerate
tok/s
Prompt
tok/s
QuantBackendTestSource
Llama 2 7B1001,705Q4_0Vulkan FA ontg128llama.cpp #10879 details
Llama 2 7B1001,753Q4_0Vulkan FA offtg128llama.cpp #10879 details
Llama 2 7B83.91,447Q4_0ROCm FA offtg128llama.cpp #15021 details
Llama 3.2 1B Instruct91.64,377Q4_K_Mllamafile LocalScore avgLocalScore details
Llama 3.1 8B Instruct37.6951Q4_K_Mllamafile LocalScore avgLocalScore details
Qwen2.5 14B Instruct21.4512Q4_K_Mllamafile LocalScore avgLocalScore details

What fits on the AMD Radeon RX 6800 XT

Estimates for this chip are scaled from a measured run: Llama 2 7B Q4_0 at 100 tok/s (Vulkan, source). Fit assumes the whole model in GPU memory with ~8K context.

ModelBest quant that fitsFileExpected tok/s
Qwen3.8-27BUD-IQ3_XXS10.9 GB35.1 estimate
gemma-4-12B-itQ8_012.7 GB30.3 estimate
Qwen3.5-9BQ8_09.5 GB40.3 estimate
Qwen3.5-4BQ8_04.5 GB85.7 estimate
Qwen3.6-35B-A3BUD-IQ3_XXS13.2 GB170 estimate
gemma-4-26B-A4B-itUD-IQ4_XS13.6 GB91.9 estimate
MiMo-V2.6-Distill-Qwen-9BQ8_09.5 GB40.2 estimate
gemma-4-E4B-itQ8_08.2 GB46.9 estimate
Muse-Glimmer-30BUD-IQ3_XXS13.1 GB29.3 estimate
Qwen3-4BQ8_04.3 GB89.8 estimate
gemma-4-31B-itUD-IQ3_XXS11.8 GB32.5 estimate
gemma-4-12B-it-qat-q4_0-unquantizedUD-Q4_K_XL6.7 GB57.2 estimate
gemma-3-1b-itQ8_01.1 GB320 estimate
gemma-4-E2B-itQ8_05.0 GB77.3 estimate

Open in the picker

Similar hardware

AMD Radeon RX 9070 XT · AMD Radeon RX 9070 · 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 9060 XT