Hardware

Radeon RX 6650 XT

Radeon RX 6650 XT has 8 GB GDDR6 and 280 GB/s. See which open models fit, at which quant, and roughly how fast. Figures below use the default memory (8 GB), 32 GB system RAM, and DDR5-5600 for anything that spills. Updated 2026-10-17.

Memory8 GB GDDR6
Bandwidth280 GB/s
Kindgpu
Year2022

128-bit GDDR6 at 17.5 Gbps = 280 GB/s. Source.

RSS for new models on this device

Models that fit

ModelVerdictQuantContexttok/s
GLM 4.6V Flash
9.4B
Tight fitQ5_K_S8K22–30
GLM Z1 9B 0414
9.4B
Tight fitQ5_K_S8K22–30
GLM 4 9B 0414
9.4B
Tight fitQ5_K_S8K22–30
GLM 4.1V 9B Thinking
9.4B
Tight fitQ5_K_S8K22–30
Qwen3.5 9B
9.0B
Tight fitUD-Q5_K_XL8K22–30
MiMo V2.6 Distill Qwen 9B
9.0B
Tight fitQ5_K_M4K22–30
gemma 4 12b it
11.9B
Tight fitIQ4_XS4K19–26
Ministral 3 8B Instruct 2512
8.5B
Tight fitUD-Q5_K_XL4K23–32
Ministral 3 8B Reasoning 2512
8.5B
Tight fitUD-Q5_K_XL4K23–32
Qwen3.8 27B
27.3B
Tight fitUD-IQ1_M4K22–30
Qwen3 VL 8B Instruct
8.2B
Tight fitUD-Q5_K_XL8K22–30
DeepSeek R1 0528 Qwen3 8B
8.2B
Tight fitUD-Q5_K_XL8K22–30
Qwen3 8B
8.2B
Tight fitUD-Q5_K_XL8K22–30
gemma 3 27b it
27.0B
Tight fitUD-IQ1_S1K22–30
Qwen3 VL 8B Thinking
8.2B
Tight fitUD-Q5_K_XL8K22–30
medgemma 27b text it
27.0B
Tight fitUD-IQ1_S2K20–28
DeepSeek R1 Distill Llama 8B
8.0B
Tight fitUD-Q5_K_XL8K23–31
Llama 3.1 8B Instruct
8.0B
Tight fitUD-Q5_K_XL8K23–31
Apertus 8B Instruct 2509
8.1B
Tight fitUD-Q5_K_XL8K22–30
Qwen2.5 7B Instruct
7.6B
RunsQ5_K_M8K26–36
Qwen2.5 VL 7B Instruct
7.6B
Tight fitQ6_K8K23–31
DeepSeek R1 Distill Qwen 7B
7.6B
Tight fitQ6_K8K23–31
Qwen2.5 Coder 7B Instruct 128K
7.6B
Tight fitQ6_K8K23–31
gemma 4 E4B it
7.5B
Tight fitUD-Q5_K_XL8K21–29
Devstral Small 2 24B Instruct 2512
23.6B
Tight fitUD-IQ1_M4K23–31
Mistral Small 3.2 24B Instruct 2506
23.6B
Tight fitUD-IQ1_M4K23–31
Magistral Small 2509
23.6B
Tight fitUD-IQ2_XXS2K22–30
Qwen2.5 14B Instruct
14.8B
Tight fitQ3_K_S2K22–30
gemma 4 E4B it qat
7.5B
RunsUD-Q4_K_XL8K31–43
Olmo 3 7B Instruct
7.3B
Tight fitUD-Q4_K_XL4K23–31
Qwen3 14B
14.8B
Tight fitUD-IQ3_XXS4K23–31
Ministral 3 14B Instruct 2512
13.5B
Tight fitQ3_K_S4K23–31
Ministral 3 14B Reasoning 2512
13.5B
Tight fitQ3_K_S4K23–31
LFM2.5 8B A1B
8.5B
Tight fitUD-Q5_K_XL8K133–181
NousResearch Hermes 4 14B
14.8B
Tight fitIQ3_XXS4K23–32
Qwen2.5 Coder 14B Instruct 128K
14.8B
Tight fitQ2_K4K23–32
gemma 4 E2B it
4.6B
RunsUD-Q8_K_XL8K28–38
Qwen3.5 4B
4.2B
Tight fitUD-Q8_K_XL8K25–34
Qwen3 4B
4.0B
Tight fitUD-Q8_K_XL8K25–34
Qwen3 VL 4B Instruct
4.0B
Tight fitUD-Q8_K_XL8K25–34

90 more models in the catalog do not fit this default profile, or ranked below the first 40 that do.

Example: Tight fit fully on your Radeon RX 6650 XT at Q5_K_S, 8K context, about 22–30 tok/s. Weights 6.24 GB plus 0.31 GB of KV cache — 6.91 GB of 7.13 GB usable. Weights shown in the verdict use GiB (the same 1024³ counting as GPU memory).