FitMyLLM
AMD· RDNA2

AMD RX 6650 XT

Running LLMs on the RX 6650 XT — the long read: which models fit at which quantisation, and the settings worth changing. · Or what else $399 buys

VRAM
8 GB
Entry-level
Bandwidth
280
GB/s
Models Q4
214/498
43%
7B Q4 speed
~36
Fast
Model coverage at Q443% of all
Estimated speed· By model size at Q4

Average speeds at Q4 quantization. Actual performance varies by model architecture and context length.

3B
~83
tok/s
7B
~36
tok/s
14B
—
7.9GB needed
32B
—
18.0GB needed
70B
—
39.4GB needed
Measured rig reports

We rent the machine and time every model on it: decode, VRAM peak, concurrency, watts and cost per million tokens. Yours may already be one of them — and two are free to read in full.

See the reports →
Specifications
VRAM
8 GB
Bandwidth
280 GB/s
FP16 Compute
21.6 TFLOPS
TDP
180W
Memory
GDDR6
Architecture
RDNA2
Stream Processors
2,048
Compute Units
32
PCIe
Gen 4 x8
MSRP
$399
201
Fast models · >30 tok/s
Real-time chat speed
214
Usable · >10 tok/s
Comfortable for all tasks
214
Total compatible
Fit in VRAM at Q4
Don’t want to buy?

Test RX 6650 XT (or anything bigger) without committing. Pay by the second, cancel anytime.

Spin up in ~60s. Pay by the second. Cancel anytime.

Some links are affiliate links — we may earn a small commission at no extra cost to you. This helps keep FitMyLLM free and independent.

Compatible models· 214
S
Falcon-H1R Tiny 90M0.09B
Falcon·256K ctx· chat· reasoning
2765
tok/s · 7% VRAM
›
S
SmolLM2 135M0.135B
Smollm·2K ctx· chat
1844
tok/s · 7% VRAM
›
S
nomic-embed-text-v1.5 100M0.14B
Embedding·8K ctx· chat
1778
tok/s · 7% VRAM
›
S
GPT-2 124M0.14B
Gpt2·1K ctx· chat
1778
tok/s · 7% VRAM
›
S
SmolVLM 256M0.256B
Smollm·8K ctx· chat· vision
972
tok/s · 8% VRAM
›
S
Gemma 3 270M0.27B
Gemma·32K ctx· chat
922
tok/s · 8% VRAM
›
S
Snowflake Arctic Embed M v2.00.305B
Embedding·8K ctx· embedding· multilingual
816
tok/s · 8% VRAM
›
S
bge-large-en-v1.5 335M0.335B
Embedding·1K ctx· chat
743
tok/s · 9% VRAM
›
S
mxbai-embed-large-v10.335B
Embedding·1K ctx· embedding
743
tok/s · 9% VRAM
›
S
Snowflake Arctic Embed L0.335B
Embedding·1K ctx· embedding
743
tok/s · 9% VRAM
›
S
LFM2 350M0.35B
Lfm·125K ctx· chat· tool_use
711
tok/s · 9% VRAM
›
S
SmolLM2 360M0.36B
Smollm·8K ctx· chat
691
tok/s · 9% VRAM
›
S
GPT-2 Medium 345M0.38B
Gpt2·1K ctx· chat
655
tok/s · 9% VRAM
›
S
Qwen 2.5 0.5B0.5B
Qwen·32K ctx· chat
498
tok/s · 10% VRAM
›
S
SmolVLM 500M0.5B
Smollm·8K ctx· chat· vision
498
tok/s · 10% VRAM
›
S
Falcon-H1 0.5B0.52B
Falcon·128K ctx· chat
479
tok/s · 10% VRAM
›
S
BGE-M30.568B
Embedding·8K ctx· embedding
438
tok/s · 10% VRAM
›
S
Snowflake Arctic Embed L v2.00.568B
Embedding·8K ctx· embedding· multilingual
438
tok/s · 10% VRAM
›
S
Qwen3-Embedding 0.6B0.6B
Embedding·32K ctx· embedding· multilingual
415
tok/s · 11% VRAM
›
S
Falcon-H1R Tiny 0.6B0.6B
Falcon·32K ctx· chat· reasoning
415
tok/s · 11% VRAM
›
S
Falcon Perception 0.6B0.6B
Falcon·4K ctx· vision
415
tok/s · 11% VRAM
›
S
Qwen 1.5 0.5B0.62B
Qwen·32K ctx· chat
401
tok/s · 11% VRAM
›
S
LFM2 700M0.74B
Lfm·125K ctx· chat· tool_use
336
tok/s · 12% VRAM
›
S
Qwen3 0.6B0.75B
Qwen·32K ctx· chat· reasoning
332
tok/s · 12% VRAM
›
S
GPT-2 Large 774M0.81B
Gpt2·1K ctx· chat
307
tok/s · 12% VRAM
›
S
Qwen 3.5 0.8B0.87B
Qwen·256K ctx· chat· coding· multilingual
286
tok/s · 13% VRAM
›
S
InternVL3 1B0.94B
Other·32K ctx· chat· vision
265
tok/s · 13% VRAM
›
S
MiniCPM5 1B1.08B
Minicpm·128K ctx· chat· reasoning· multilingual
230
tok/s · 14% VRAM
›
S
K2 Horizon 0.9B1.08B
Other·128K ctx· chat· coding· reasoning
230
tok/s · 14% VRAM
›
S
TinyLlama 1.1B1.1B
Llama·2K ctx· chat
226
tok/s · 15% VRAM
›
S
LFM2.5-1.2B-Thinking1.2B
Lfm·122K ctx· chat· reasoning· tool_use
207
tok/s · 15% VRAM
›
S
Llama-3.2-1B1.2B
Llama·4K ctx· chat
207
tok/s · 15% VRAM
›
S
LFM2 1.2B1.2B
Lfm·125K ctx· chat· tool_use· multilingual
207
tok/s · 15% VRAM
›
S
Zamba2 1.2B1.2B
Other·4K ctx· chat
207
tok/s · 15% VRAM
›
S
Ling 3.0 Tiny7.89BMoE
Other·128K ctx· chat· coding· reasoning
191
tok/s · 66% VRAM
›
S
EXAONE-4.0-1.2B1.3B
Exaone·64K ctx· chat
191
tok/s · 16% VRAM
›
S
OPT 1.3B1.3B
Opt·2K ctx· chat
191
tok/s · 16% VRAM
›
S
MiniCPM-V 4.61.3B
Other·256K ctx· chat· vision
191
tok/s · 16% VRAM
›
S
DeepSeek Coder 1.3B1.35B
Deepseek·16K ctx· coding
184
tok/s · 16% VRAM
›
S
Phi-1 1.3B1.42B
Phi·2K ctx· coding
175
tok/s · 17% VRAM
›
Next step

Get personalized recommendations.

See ranked models with benchmark scores, run commands, and precise speed estimates for your RX 6650 XT.

What this card is worth

RX 6650 XT holds 214 of the models in our catalogue and is, in practice, a IQ4_XS card — the largest it takes is Falcon3-10B at IQ4_XS.

Tokens/sec per $100
7.8
8B at Q4_K_M, so cards compare like for like
VRAM per $100
2.01 GB
what memory costs on this card
The biggest it takes
Falcon3-10B10.3B · IQ4_XS7.1 GB26 tok/sest
Bamba 9B v29.78B · Q4_K_M7.2 GB25 tok/sest
RecurrentGemma 9B9.63B · Q4_K_M6.8 GB25 tok/sest
glm-4-9b9.4B · Q4_K_M6.6 GB26 tok/sest
Yi 1.5 9B9B · Q4_K_M6.5 GB27 tok/sest
Yi Coder 9B9B · Q4_K_M6.5 GB27 tok/sest
K2 Horizon 7B9B · Q4_K_M6.7 GB27 tok/sest
Ministral 3 8B8.92B · Q4_K_M6.6 GB27 tok/sest
Against cards of similar memory
M3 Pro (18GB)12 GB$1,5991.5 tok/s per $100
M1 Pro (16GB)11 GB$9992.9 tok/s per $100
M2 Pro (16GB)11 GB$1,2992.2 tok/s per $100
M4 (16GB)11 GB$4994.0 tok/s per $100
RX 6650 XT8 GB$3997.8 tok/s per $100

Neighbours in memory rather than in price: memory decides whether a card can do the job at all, so two cards of the same size at different prices is the comparison you are making. Ordered by memory, then bandwidth — not by the value column, which is worked out from bandwidth and price alone and therefore rewards a cheap card whatever its software stack does to that bandwidth in practice. Read it as one input, not as a ranking.

Device under test

AMD RX 6650 XT — 8 GB VRAM.

RX 6650 XT spec
Brand
AMD
VRAM
8 GB GDDR6
Bandwidth
280 GB/s
FP16 Compute
21.6 TFLOPS
FP32 Compute
10.8 TFLOPS
Stream Processors
2,048
TDP
180 W
Architecture
RDNA2
MSRP
$399
AI capability
214/ 498 models @ Q4

With 8 GB VRAM and 280 GB/s bandwidth, this GPU handles models up to 9.63B parameters.

Speed ≈ bandwidth / model_size × efficiency. A 7B model at Q4 runs at ~32 tok/s.

Rent it — live prices
Checking what the clouds are charging…
Top models for RX 6650 XT
214 fit · showing 20
ModelSizeVRAM Q4tok/sAvg
RecurrentGemma 9B9.63B6.4 GB2635.0
glm-4-9b9.4B6.2 GB2620.5
gemma-2-9b9.2B6.1 GB27171.8
Yi 1.5 9B9B6.0 GB2830.3
Yi Coder 9B9B6.0 GB2835.8
K2 Horizon 7B9B6.0 GB2852.4
Ministral 3 8B8.92B5.9 GB2822.6
Ministral 3 8B Reasoning8.92B5.9 GB2869.0
NVIDIA-Nemotron-Nano-9B-v28.9B5.9 GB2844.2
InternLM3 8B Instruct8.8B5.9 GB2842.8
Granite 4.2 8B8.79B5.9 GB2844.9
Gemma 1 7B8.54B5.7 GB2924.7
CodeGemma 7B8.54B5.7 GB2940.2
LFM2 8B A1B8.3B5.6 GB16624.3
Seed-Coder 8B Instruct8.25B5.5 GB3034.1
Seed-Coder 8B Reasoning8.25B5.5 GB3032.9
DeepSeek R1-0528 Qwen3 8B8.2B5.5 GB3036.3
Qwen3-8B8.2B5.5 GB3086.5
Granite 3.0 8B8.17B5.5 GB3036.4
Granite 3.1 8B8.17B5.5 GB3038.6