Skip to content

NVIDIA GeForce GTX 1660 Super

NVIDIA · 6GB GDDR6 · Can run 26 models

Buy Amazon
Manufacturer NVIDIA
VRAM 6 GB
Memory Type GDDR6
Architecture Turing
CUDA Cores 1,408
Bandwidth 336 GB/s
TDP 125W
MSRP $229
Released Oct 29, 2019

AI Notes

The GTX 1660 Super's 6GB VRAM limits it to smaller 7B models with tight quantization (Q4 or lower). It lacks tensor cores, so inference speed relies entirely on CUDA throughput. Widely available used at low prices, it's a minimal entry point for experimenting with local AI on a tight budget.

Compatible Models

Model Parameters Best Quant VRAM Used Fit Est. Speed
Qwen 3 0.6B 600M Q4_K_M 2.5 GB Runs ~134 tok/s
Gemma 3 1B 1B Q8_0 2 GB Runs ~168 tok/s
Llama 3.2 1B 1B Q8_0 3 GB Runs ~112 tok/s
DeepSeek R1 1.5B 1.5B Q8_0 3 GB Runs ~112 tok/s
Gemma 2 2B 2B Q8_0 4 GB Runs ~84 tok/s
Gemma 3n E2B 2B Q4_K_M 3.3 GB Runs ~102 tok/s
Llama 3.2 3B 3B Q8_0 5 GB Runs ~67 tok/s
Phi-4 Mini 3.8B 3.8B Q4_K_M 4.5 GB Runs ~75 tok/s
Gemma 3 4B 4B Q4_K_M 5 GB Runs ~67 tok/s
Gemma 3n E4B 4B Q4_K_M 4.5 GB Runs ~75 tok/s
Qwen 3 4B 4B Q4_K_M 4.5 GB Runs ~75 tok/s
Phi-3 Mini 3.8B 3.8B Q8_0 5.8 GB CPU Offload ~17 tok/s
DeepSeek R1 7B 7B Q8_0 9 GB CPU Offload ~11 tok/s
Falcon 3 7B 7B Q4_K_M 6.8 GB CPU Offload ~15 tok/s
Mistral 7B 7B Q8_0 9 GB CPU Offload ~11 tok/s
Qwen 2.5 7B 7B Q8_0 9 GB CPU Offload ~11 tok/s
Qwen 2.5 Coder 7B 7B Q8_0 9 GB CPU Offload ~11 tok/s
Qwen 2.5 VL 7B 7B Q4_K_M 7 GB CPU Offload ~14 tok/s
Cogito 8B 8B Q4_K_M 7.5 GB CPU Offload ~14 tok/s
DeepSeek R1 8B 8B Q4_K_M 7.5 GB CPU Offload ~14 tok/s
Llama 3.1 8B 8B Q4_K_M 6.3 GB CPU Offload ~16 tok/s
Nemotron 3 Nano 8B 8B Q4_K_M 7.5 GB CPU Offload ~14 tok/s
Qwen 3 8B 8B Q4_K_M 7.5 GB CPU Offload ~14 tok/s
Gemma 2 9B 9B Q4_K_M 6.9 GB CPU Offload ~15 tok/s
Falcon 3 10B 10B Q4_K_M 8.5 GB CPU Offload ~12 tok/s
Llama 3.2 Vision 11B 11B Q4_K_M 8.5 GB CPU Offload ~12 tok/s
43 model(s) are too large for this hardware.