Skip to content

AMD Radeon RX 7700 XT

AMD · 12GBGDDR6 · Can run 63 models

BuyAmazon
ManufacturerAMD
VRAM12 GB
Memory TypeGDDR6
ArchitectureRDNA 3
Stream Procs3,456
Bandwidth432 GB/s
TDP245W
MSRP$449
ReleasedSep 6, 2023

AI Notes

The RX 7700 XT offers 12GB VRAM at a competitive price, handling 7B models well and running 13B with Q4 quantization. The 432 GB/s bandwidth provides moderate inference speeds. ROCm support requires additional setup compared to NVIDIA, but the extra VRAM over 8GB cards opens up more model options.

Compatible Models

ModelParametersBest QuantVRAM UsedFitEst. Speed
Qwen 3 0.6B600MQ4_K_M2.5 GBRuns~173 tok/s
Qwen 3.5 0.8B800MQ4_K_M1.5 GBRuns~288 tok/s
Gemma 3 1B1BQ8_02 GBRuns~216 tok/s
Llama 3.2 1B1BQ8_03 GBRuns~144 tok/s
DeepSeek R1 1.5B1.5BQ8_03 GBRuns~144 tok/s
SmolLM2 1.7B1.7BQ8_02.7 GBRuns~160 tok/s
Gemma 2 2B2BQ8_04 GBRuns~108 tok/s
Gemma 3n E2B2BQ4_K_M3.3 GBRuns~131 tok/s
Gemma 4 E2B2BQ4_K_M4 GBRuns~108 tok/s
Qwen 3.5 2B2BQ4_K_M3 GBRuns~144 tok/s
Llama 3.2 3B3BQ8_05 GBRuns~86 tok/s
StarCoder2 3B3BQ4_K_M3.5 GBRuns~123 tok/s
Phi-3 Mini 3.8B3.8BQ8_05.8 GBRuns~74 tok/s
Phi-4 Mini 3.8B3.8BQ4_K_M4.5 GBRuns~96 tok/s
Gemma 3 4B4BQ4_K_M5 GBRuns~86 tok/s
Gemma 3n E4B4BQ4_K_M4.5 GBRuns~96 tok/s
Gemma 4 E4B4BQ4_K_M6 GBRuns~72 tok/s
Qwen 3 4B4BQ4_K_M4.5 GBRuns~96 tok/s
Qwen 3.5 4B4BQ4_K_M4.5 GBRuns~96 tok/s
Yi 1.5 6B6BQ4_K_M5 GBRuns~86 tok/s
Codestral Mamba 7B7BQ4_K_M6.9 GBRuns~63 tok/s
DeepSeek R1 7B7BQ8_09 GBRuns~48 tok/s
Falcon 3 7B7BQ4_K_M6.8 GBRuns~64 tok/s
InternLM 2.5 7B7BQ4_K_M5.5 GBRuns~79 tok/s
Mistral 7B7BQ8_09 GBRuns~48 tok/s
OpenChat 3.5 7B7BQ4_K_M6.9 GBRuns~63 tok/s
Qwen 2.5 7B7BQ8_09 GBRuns~48 tok/s
Qwen 2.5 Coder 7B7BQ8_09 GBRuns~48 tok/s
Qwen 2.5 VL 7B7BQ4_K_M7 GBRuns~62 tok/s
StarCoder2 7B7BQ4_K_M5.5 GBRuns~79 tok/s
WizardLM 2 7B7BQ4_K_M6.9 GBRuns~63 tok/s
Aya Expanse 8B8BQ4_K_M6.5 GBRuns~66 tok/s
Cogito 8B8BQ4_K_M7.5 GBRuns~58 tok/s
DeepSeek R1 8B8BQ4_K_M7.5 GBRuns~58 tok/s
Dolphin 3 8B8BQ4_K_M6 GBRuns~72 tok/s
Granite 3.3 8B8BQ8_010 GBRuns~43 tok/s
Llama 3.1 8B8BQ8_010 GBRuns~43 tok/s
Nemotron 3 Nano 8B8BQ4_K_M7.5 GBRuns~58 tok/s
Nous Hermes 2 8B8BQ4_K_M6 GBRuns~72 tok/s
Qwen 3 8B8BQ4_K_M7.5 GBRuns~58 tok/s
Qwen 3.5 9B9BQ4_K_M7.5 GBRuns~58 tok/s
Yi 1.5 9B9BQ4_K_M6.5 GBRuns~66 tok/s
Yi Coder 9B9BQ4_K_M8 GBRuns~54 tok/s
Falcon 3 10B10BQ4_K_M8.5 GBRuns~51 tok/s
Llama 3.2 Vision 11B11BQ4_K_M8.5 GBRuns~51 tok/s
Mistral Nemo 12B12BQ4_K_M9.5 GBRuns~45 tok/s
DeepSeek R1 14B14BQ4_K_M9.9 GBRuns~44 tok/s
Phi-4 14B14BQ4_K_M9.9 GBRuns~44 tok/s
Qwen 2.5 14B14BQ4_K_M9.9 GBRuns~44 tok/s
Gemma 2 9B9BQ8_011 GBRuns (tight)~39 tok/s
Gemma 3 12B12BQ4_K_M10.5 GBRuns (tight)~41 tok/s
Phi-4 Reasoning 14B14BQ4_K_M11 GBRuns (tight)~39 tok/s
Qwen 2.5 Coder 14B14BQ4_K_M12 GBCPU Offload~11 tok/s
Qwen 3 14B14BQ4_K_M12 GBCPU Offload~11 tok/s
StarCoder2 15B15BQ8_017 GBCPU Offload~8 tok/s
InternLM 2.5 20B20BQ4_K_M12 GBCPU Offload~11 tok/s
gpt-oss 20B21BMXFP415 GBCPU Offload~9 tok/s
Codestral 22B22BQ4_K_M14.7 GBCPU Offload~9 tok/s
Devstral 24B24BQ4_K_M17 GBCPU Offload~8 tok/s
Magistral Small 24B24BQ4_K_M17 GBCPU Offload~8 tok/s
Mistral Small 3.1 24B24BQ4_K_M18 GBCPU Offload~7 tok/s
Gemma 2 27B27BQ4_K_M17.7 GBCPU Offload~7 tok/s
Qwen 3.5 35B A3B35BQ4_K_M12 GBCPU Offload~11 tok/s
51 model(s) are too large for this hardware.