Hardware · GPU · NVIDIA

NVIDIA GeForce RTX 4060 Ti 16GB

16 GB of GDDR6 memory at 288 GB/s — enough for models up to ~22B parameters at 4-bit on one card.

What it runs

Largest models without offloading on the RTX 4060 Ti 16GB budget build (32 GB DDR5), 8K context.

Variant
Qwen2.5 14B Instruct
14.8B · Q6_K via llama.cpp
Tight fit
~15 tok/s est.
Variant
Phi-4
14.7B · Q4_K_M via llama.cpp
Runs well
~20 tok/s est.
Variant
Gemma 2 9B IT
9.24B · Q4_K_M via llama.cpp
Runs well
~30 tok/s est.
Variant
Hermes 3 Llama 3.1 8B
8.03B · Q8_0 via llama.cpp
Runs well
~21 tok/s est.
Variant
Llama 3.1 8B Instruct
8.03B · Q8_0 via llama.cpp
Runs well
~21 tok/s est.
Variant
Qwen2.5 7B Instruct
7.62B · Q8_0 via llama.cpp
Runs well
~22 tok/s est.
Variant
DeepSeek-R1-Distill-Qwen-7B
7.62B · Q4_K_M via llama.cpp
Runs well
~36 tok/s est.
Variant
Mistral 7B Instruct v0.3
7.25B · Q8_0 via llama.cpp
Runs well
~23 tok/s est.

System RTX 4060 Ti 16GB budget build (32 GB DDR5)
Runs models up to ~22B parameters at 4-bit entirely on the GPU. What runs

Throughput on systems containing this device.

Model · quantization	System	Runtime	Context	Measurements	Source
Qwen2.5 14B Instruct Q4_K_M	RTX 4060 Ti 16GB budget build (32 GB DDR5)	llama.cppcuda · b4600	4K	1,250 tok/s · Prompt processing (512) 25.5 tok/s · Generation (128)	Mutinai illustrative fixtures

Contributions are not open yet, so there is nothing here from members.

Contributions are not open yet, so there is nothing here from members.

No source records linked.