GPUs & Macs › AMD Radeon RX 7900 XT

AMD Radeon RX 7900 XT: local LLM speed

Llama 2 7B (Q4_0) generates 123 tokens/s on the AMD Radeon RX 7900 XT (Vulkan), and processes prompts at 2,942 tokens/s.

Measured results

ModelGenerate
tok/s
Prompt
tok/s
QuantBackendTestSource
Llama 2 7B1232,942Q4_0Vulkan FA offtg128llama.cpp #10879 details
Llama 2 7B1212,701Q4_0Vulkan FA ontg128llama.cpp #10879 details
Llama 2 7B1163,098Q4_0ROCm FA offtg128llama.cpp #15021 details
Llama 2 7B1123,262Q4_0ROCm FA ontg128llama.cpp #15021 details
gpt-oss 20B1024,252BF16ROCm,RPC FA ontg128llama.cpp #15396 details
Llama 3.2 1B Instruct99.24,957Q4_K_Mllamafile LocalScore avgLocalScore details

What fits on the AMD Radeon RX 7900 XT

Estimates for this chip are scaled from a measured run: Llama 2 7B Q4_0 at 123 tok/s (Vulkan, source). Fit assumes the whole model in GPU memory with ~8K context.

ModelBest quant that fitsFileExpected tok/s
Qwen3.8-27BUD-Q4_K_M16.5 GB28.6 estimate
gemma-4-12B-itQ8_012.7 GB37.2 estimate
Qwen3.5-9BQ8_09.5 GB49.4 estimate
Qwen3.5-4BQ8_04.5 GB105 estimate
Qwen3.6-35B-A3BUD-IQ4_XS17.7 GB155 estimate
gemma-4-26B-A4B-itUD-Q4_K_M16.9 GB90.3 estimate
MiMo-V2.6-Distill-Qwen-9BQ8_09.5 GB49.3 estimate
gemma-4-E4B-itQ8_08.2 GB57.5 estimate
Muse-Glimmer-30BUD-IQ3_M14.1 GB33.3 estimate
Qwen3-4BQ8_04.3 GB110 estimate
gemma-4-31B-itQ4_K_S17.4 GB27.1 estimate
gemma-4-26B-A4B-it-qat-q4_0-unquantizedUD-Q4_K_XL14.2 GB107 estimate
gemma-4-12B-it-qat-q4_0-unquantizedUD-Q4_K_XL6.7 GB70.1 estimate
gemma-3-1b-itQ8_01.1 GB392 estimate

Open in the picker

Similar hardware

AMD Radeon RX 7900 XTX · 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