Which local LLMs run on Radeon RX 7900 XT?

On this site, 912 models have a quantization that fits in 20 GB at 8,192 tokens (estimated). Popular ones include Qwen3.8-27B, Ornith-1.5-9B, Huihui-Qwen3.6-35B-A3B-abliterated-MTP, Ornith-1.5-35B-A3B, Huihui-Qwen3.8-27B-abliterated.

What is the largest model that runs on Radeon RX 7900 XT?

Among the models listed here, Ling-3.0-flash (about 127.5B); Q1_0 fits in 20 GB (estimated at 8,192 tokens).

How fast does Radeon RX 7900 XT generate text?

With 800 GB/s of memory bandwidth, Qwen3.8-27B at Q5_K_S (19.6 GB) runs at an estimated 25–33 tokens per second (60–80% of bandwidth ÷ file size; prompt processing not included).

Discrete GPU

Radeon RX 7900 XT

Data as of Nov 11, 2026

Memory bandwidth: 800 GB/s
Source: Checked: Sep 27, 2026

With 20 GB

One entry per model, up to 30, sorted by popularity. For each model we show the largest (most precise) quantization that fits in this memory with a context of at least 8K tokens. The context length is the longest one that quantization fits (for models whose maximum is under 8K, such as embedding models, that maximum). Models that are not for chat or text generation are labelled with their use. "Left" is the memory minus the estimate; under 5% of the memory is marked "tight" (not a guarantee that it fits).

Qwen3.8-27B Q5_K_S Image input

Context length 8,192 Estimated 21.3 GB 0.17 GB left (tight) Speed 25–33
Ornith-1.5-9B BF16

Context length 32,768 Estimated 20.6 GB 0.87 GB left (tight) Speed 26–35
Huihui-Qwen3.6-35B-A3B-abliterated-MTP Q3_K Image input

Context length 131,072 Estimated 21.0 GB 0.51 GB left (tight) Speed not estimated (MoE)
Huihui-Qwen3.8-27B-abliterated Q5_K_M

Context length 8,192 Estimated 21.3 GB 0.20 GB left (tight) Speed 25–33
OTel-2.0-Decision-26B-A4B-4bit Q4_K_M

Context length 131,072 Estimated 20.9 GB 0.61 GB left (tight) Speed not estimated (MoE)
Qwen3.8-27B-iMatrix-NVFP4-MTP NVFP4 Image input

Context length 32,768 Estimated 20.5 GB 1.00 GB left (tight) Speed 28–37
Tiel-Coder-35B-A3B-GGUF-MTP UD-IQ4_XS Image input

Context length 65,536 Estimated 20.6 GB 0.90 GB left (tight) Speed not estimated (MoE)
gemma-4-12B-it UD-Q8_K_XL Image input

Context length 262,144 Estimated 19.5 GB 1.97 GB left Speed 35–47
Qwen3.8-27B-Uncensored Q5_K_M

Context length 8,192 Estimated 21.3 GB 0.20 GB left (tight) Speed 25–33
Ornith-1.0-9B BF16

Context length 65,536 Estimated 21.2 GB 0.28 GB left (tight) Speed 27–36
mxbai-embed-large-v1 F32 Embedding (not for chat)

Context length 512 Estimated 2.41 GB 19.1 GB left Speed 359–479
Qwen3.6-35B-A3B-MTP UD-IQ4_NL Image input

Context length 65,536 Estimated 21.0 GB 0.49 GB left (tight) Speed not estimated (MoE)
Qwen3.5-9B-The-Defiant-Fable-Uncensored-Heretic-NEO-IMATRIX-MAX-MTP BF16 Image input

Context length 32,768 Estimated 20.6 GB 0.90 GB left (tight) Speed 26–35
MOSS-Music-8B-Instruct F16 Speech / audio

Context length 16,384 Estimated 19.7 GB 1.82 GB left Speed 29–39
Qwen3.8-27B-Uncensored-HauhauCS-Aggressive-MTP Q4_K Image input

Context length 32,768 Estimated 21.3 GB 0.20 GB left (tight) Speed 27–36
gemma-4-26B-A4B-it Q5_K_L Image input

Context length 16,384 Estimated 21.2 GB 0.25 GB left (tight) Speed not estimated (MoE)
Qwen3-VL-8B-Instruct BF16 Image input

Context length 16,384 Estimated 19.7 GB 1.82 GB left Speed 29–39
Qwen3.5-9B BF16 Image input

Context length 32,768 Estimated 20.6 GB 0.90 GB left (tight) Speed 26–35
Ornith-1.5-9B-uncensored F16

Context length 65,536 Estimated 21.2 GB 0.31 GB left (tight) Speed 27–36
Qwen3.6-35B-A3B IQ4_XS Image input

Context length 16,384 Estimated 21.1 GB 0.33 GB left (tight) Speed not estimated (MoE)
Ternary-Bonsai-2-27B Q2_0

Context length 131,072 Estimated 17.4 GB 4.06 GB left Speed 63–84
Swift-1.5-Qwen3.8-27B-GSQ-RCO IQ3_S

Context length 65,536 Estimated 17.6 GB 3.86 GB left Speed 40–53
Qwen3.5-4B BF16 Image input

Context length 262,144 Estimated 18.4 GB 3.12 GB left Speed 55–74
Qwen3.8-27B-TURBO-Fable-Cold-Fusion-735-882-Heretic-Uncensored-NEO-CODER-MAX-MTP Q4_K_M Image input

Context length 16,384 Estimated 20.8 GB 0.70 GB left (tight) Speed 26–35
Dirk-Qwen3.8-27B UD-Q4_K_XL Image input

Context length 32,768 Estimated 20.9 GB 0.57 GB left (tight) Speed 27–36
Wan2.1 Q4_K_M

Context length 32,768 Estimated 20.2 GB 1.29 GB left Speed 29–38
Qwen3.8-27B-GSQ-RCO IQ3_S Image input

Context length 65,536 Estimated 17.6 GB 3.86 GB left Speed 40–53
Qwen3-ASR-1.7B F16 Speech / audio

Context length 65,536 Estimated 12.5 GB 9.02 GB left Speed 117–156
Qwen3.8-35B-A3B-Distill IQ4_XS

Context length 32,768 Estimated 21.4 GB 0.06 GB left (tight) Speed not estimated (MoE)
Qwen3.6-35B-A3B-Uncensored-HauhauCS-Aggressive IQ4_NL Image input

Context length 16,384 Estimated 21.2 GB 0.25 GB left (tight) Speed not estimated (MoE)

Frequently asked questions

Which local LLMs run on Radeon RX 7900 XT?: On this site, 912 models have a quantization that fits in 20 GB at 8,192 tokens (estimated). Popular ones include Qwen3.8-27B, Ornith-1.5-9B, Huihui-Qwen3.6-35B-A3B-abliterated-MTP, Ornith-1.5-35B-A3B, Huihui-Qwen3.8-27B-abliterated.
What is the largest model that runs on Radeon RX 7900 XT?: Among the models listed here, Ling-3.0-flash (about 127.5B); Q1_0 fits in 20 GB (estimated at 8,192 tokens).
How fast does Radeon RX 7900 XT generate text?: With 800 GB/s of memory bandwidth, Qwen3.8-27B at Q5_K_S (19.6 GB) runs at an estimated 25–33 tokens per second (60–80% of bandwidth ÷ file size; prompt processing not included).