Hardware / AMD /Radeon RX 7900 XTX

AMD · GPU

Radeon RX 7900 XTX

Runs 108 models fully in memory, 8 more with CPU offload. 7 of those sit close enough to 24 GB that our size estimate is what decides them.

Memory

24 GB

Bandwidth

960 GB/s

Runs

108

01

Specifications

Memory

24 GB

Bandwidth

960 GB/s

FP16 compute

122.8 TFLOPS

TDP

355 W

MSRP

$999

Street price

—

Released

Dec 13, 2022

Architecture

RDNA 3

02

What it runs

Best quant per model — sorted by fit score.

ModelParamsBest quantContextSpeedVerdict

Ministral 3 3B 25123.8BQ4_K_M131K271 tok/sestFits in memory Granite 4.0 Micro3.2BQ4_K_M66K273 tok/sestFits in memory UI-TARS 7B8.3BQ4_K_M66K124 tok/sestFits in memory Llama 3.2 3B Instruct3.2BQ4_K_M131K322 tok/sestFits in memory Qwen3 8B8.2BQ4_K_M66K126 tok/sestFits in memory Gemma 3 4B4.3BQ5_K_M131K204 tok/sestFits in memory Llama 3.1 8B Instruct8BQ4_K_M131K129 tok/sestFits in memory Niagara 19m Batch.en20MQ8_0~262K29365 tok/sestFits in memory Niagara 38m Batch.en38MQ5_K_M~262K23088 tok/sestFits in memory ARK ASR 0.6B1.3BQ5_K_M~262K675 tok/sestFits in memory ARK ASR 3B4.1BQ5_K_M~262K214 tok/sestFits in memory Audio8 ASR 0.1B0.3BQ5_K_M~262K2924 tok/sestFits in memory Higgs Audio v3 STT2.7BQ4_K_M~131K381 tok/sestFits in memory Cohere Transcribe 03 20262.1BQ8_0~262K280 tok/sestFits in memory Distil Large v3.50.8BQ8_0~262K734 tok/sestFits in memory Lite Whisper Large v3 Acc1.4BQ8_0~262K420 tok/sestFits in memory Owsm CTC v3.1 1B1.1BQ5_K_M~262K798 tok/sestFits in memory Owsm CTC v3.2 ft 1B1BQ8_0~262K587 tok/sestFits in memory Owsm CTC v4 1B1BQ8_0~262K587 tok/sestFits in memory Hubert Large Ls960 ft0.3BQ4_K_M~262K3217 tok/sestFits in memory MMS 1b All1BQ8_0~262K587 tok/sestFits in memory Wav2vec2 Large 960h Lv60 Self0.3BQ5_K_M~262K2742 tok/sestFits in memory Hojo ASR V15.2BQ5_K_M~262K169 tok/sestFits in memory Granite 4.0 1b Speech2.3BQ8_0~131K255 tok/sestFits in memory Granite Speech 3.3 2b3BQ4_K_M~131K343 tok/sestFits in memory Granite Speech 3.3 8b8.6BQ4_K_M~66K120 tok/sestFits in memory Granite Speech 4.1 2b2.3BQ4_K_M~131K448 tok/sestFits in memory Granite Speech 4.1 2b NAR2.3BQ8_0~131K255 tok/sestFits in memory STT 2.6b en2.6BQ4_K_M~262K396 tok/sestFits in memory Phi 4 Multimodal Instruct5.6BQ4_K_M~131K184 tok/sestFits in memory VibeVoice ASR HF8.3BQ4_K_M~262K124 tok/sestFits in memory Voxtral Mini 3B 25075BQ5_K_M~262K176 tok/sestFits in memory Voxtral Mini 4B Realtime 26024.4BQ4_K_M~131K234 tok/sestFits in memory Canary 180m Flash0.2BQ4_K_M~262K5718 tok/sestFits in memory Canary 1b1BQ5_K_M~262K877 tok/sestFits in memory Canary 1b Flash1BQ4_K_M~262K1029 tok/sestFits in memory Canary 1b v21BQ4_K_M~262K1029 tok/sestFits in memory Canary Qwen 2.5b2.6BQ4_K_M~262K396 tok/sestFits in memory Nemotron 3.5 ASR Streaming 0.6b0.6BQ8_0~262K918 tok/sestFits in memory Nemotron Speech Streaming en 0.6b0.6BQ8_0~262K979 tok/sestFits in memory Parakeet CTC 0.6b0.6BQ5_K_M~262K1462 tok/sestFits in memory Parakeet CTC 1.1b1.1BQ5_K_M~262K798 tok/sestFits in memory Parakeet RNNT 0.6b0.6BQ4_K_M~262K1716 tok/sestFits in memory Parakeet RNNT 1.1b1.1BQ5_K_M~262K798 tok/sestFits in memory Parakeet TDT CTC 110m0.1BQ8_0~262K5339 tok/sestFits in memory Parakeet TDT 0.6b v20.6BQ5_K_M~262K1462 tok/sestFits in memory Parakeet TDT 0.6b v30.6BQ4_K_M~262K1716 tok/sestFits in memory Parakeet TDT 1.1b1.1BQ5_K_M~262K798 tok/sestFits in memory Nyra Health CrisperWhisper1.6BQ5_K_M~262K548 tok/sestFits in memory Whisper Large v31.5BQ5_K_M~262K585 tok/sestFits in memory Whisper Large v3 Turbo0.8BQ8_0~262K734 tok/sestFits in memory MOSS Transcribe Diarize0.9BQ8_0~131K653 tok/sestFits in memory Qwen3 ASR 0.6B0.8BQ5_K_M~262K1125 tok/sestFits in memory Qwen3 ASR 0.6B HF0.8BQ4_K_M~131K1287 tok/sestFits in memory Qwen3 ASR 1.7B2BQ8_0~262K288 tok/sestFits in memory Qwen3 ASR 1.7B HF2BQ4_K_M~262K505 tok/sestFits in memory Zipformer cr CTC Transducer XL 290M0.3BQ4_K_M~262K3549 tok/sestFits in memory Zipformer Transducer XL 290M0.3BQ8_0~262K2025 tok/sestFits in memory ASR Conformer Largescaleasr0.5BQ5_K_M~262K1828 tok/sestFits in memory Moonshine Streaming Medium0.2BQ5_K_M~262K3656 tok/sestFits in memory

A speed marked est is calculated from this card's memory bandwidth and the size of the weights it has to read per word. It is not a measurement.

Something wrong on this page? Tell us