FitMyLLM
NVIDIA· Turing

NVIDIA GeForce GTX 1660 Ti Max-Q

Running LLMs on the GeForce GTX 1660 Ti Max-Q — the long read: which models fit at which quantisation, and the settings worth changing. · Or what a budget buys

VRAM
6 GB
Budget
Bandwidth
288
GB/s
Models Q4
139/498
28%
7B Q4 speed
~33
Fast
Model coverage at Q428% of all
Estimated speed· By model size at Q4

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

3B
~77
tok/s
7B
~33
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
6 GB
Bandwidth
288 GB/s
FP16 Compute
8.2 TFLOPS
TDP
60W
Memory
GDDR6
Architecture
Turing
CUDA Cores
1,536
PCIe
Gen 3 x16
139
Fast models · >30 tok/s
Real-time chat speed
139
Usable · >10 tok/s
Comfortable for all tasks
139
Total compatible
Fit in VRAM at Q4
Don’t want to buy?

Test GeForce GTX 1660 Ti Max-Q (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· 139
S
Falcon-H1R Tiny 90M0.09B
Falcon·256K ctx· chat· reasoning
2560
tok/s · 9% VRAM
›
S
SmolLM2 135M0.135B
Smollm·2K ctx· chat
1707
tok/s · 10% VRAM
›
S
nomic-embed-text-v1.5 100M0.14B
Embedding·8K ctx· chat
1646
tok/s · 10% VRAM
›
S
GPT-2 124M0.14B
Gpt2·1K ctx· chat
1646
tok/s · 10% VRAM
›
S
SmolVLM 256M0.256B
Smollm·8K ctx· chat· vision
900
tok/s · 11% VRAM
›
S
Gemma 3 270M0.27B
Gemma·32K ctx· chat
853
tok/s · 11% VRAM
›
S
Snowflake Arctic Embed M v2.00.305B
Embedding·8K ctx· embedding· multilingual
755
tok/s · 11% VRAM
›
S
bge-large-en-v1.5 335M0.335B
Embedding·1K ctx· chat
688
tok/s · 12% VRAM
›
S
mxbai-embed-large-v10.335B
Embedding·1K ctx· embedding
688
tok/s · 12% VRAM
›
S
Snowflake Arctic Embed L0.335B
Embedding·1K ctx· embedding
688
tok/s · 12% VRAM
›
S
LFM2 350M0.35B
Lfm·125K ctx· chat· tool_use
658
tok/s · 12% VRAM
›
S
SmolLM2 360M0.36B
Smollm·8K ctx· chat
640
tok/s · 12% VRAM
›
S
GPT-2 Medium 345M0.38B
Gpt2·1K ctx· chat
606
tok/s · 12% VRAM
›
S
Qwen 2.5 0.5B0.5B
Qwen·32K ctx· chat
461
tok/s · 13% VRAM
›
S
SmolVLM 500M0.5B
Smollm·8K ctx· chat· vision
461
tok/s · 13% VRAM
›
S
Falcon-H1 0.5B0.52B
Falcon·128K ctx· chat
443
tok/s · 13% VRAM
›
S
BGE-M30.568B
Embedding·8K ctx· embedding
406
tok/s · 14% VRAM
›
S
Snowflake Arctic Embed L v2.00.568B
Embedding·8K ctx· embedding· multilingual
406
tok/s · 14% VRAM
›
S
Qwen3-Embedding 0.6B0.6B
Embedding·32K ctx· embedding· multilingual
384
tok/s · 14% VRAM
›
S
Falcon-H1R Tiny 0.6B0.6B
Falcon·32K ctx· chat· reasoning
384
tok/s · 14% VRAM
›
S
Falcon Perception 0.6B0.6B
Falcon·4K ctx· vision
384
tok/s · 14% VRAM
›
S
Qwen 1.5 0.5B0.62B
Qwen·32K ctx· chat
372
tok/s · 14% VRAM
›
S
LFM2 700M0.74B
Lfm·125K ctx· chat· tool_use
311
tok/s · 16% VRAM
›
S
Qwen3 0.6B0.75B
Qwen·32K ctx· chat· reasoning
307
tok/s · 16% VRAM
›
S
GPT-2 Large 774M0.81B
Gpt2·1K ctx· chat
284
tok/s · 16% VRAM
›
S
Qwen 3.5 0.8B0.87B
Qwen·256K ctx· chat· coding· multilingual
265
tok/s · 17% VRAM
›
S
InternVL3 1B0.94B
Other·32K ctx· chat· vision
245
tok/s · 18% VRAM
›
S
MiniCPM5 1B1.08B
Minicpm·128K ctx· chat· reasoning· multilingual
213
tok/s · 19% VRAM
›
S
K2 Horizon 0.9B1.08B
Other·128K ctx· chat· coding· reasoning
213
tok/s · 19% VRAM
›
S
TinyLlama 1.1B1.1B
Llama·2K ctx· chat
209
tok/s · 19% VRAM
›
S
LFM2.5-1.2B-Thinking1.2B
Lfm·122K ctx· chat· reasoning· tool_use
192
tok/s · 20% VRAM
›
S
Llama-3.2-1B1.2B
Llama·4K ctx· chat
192
tok/s · 20% VRAM
›
S
LFM2 1.2B1.2B
Lfm·125K ctx· chat· tool_use· multilingual
192
tok/s · 20% VRAM
›
S
Zamba2 1.2B1.2B
Other·4K ctx· chat
192
tok/s · 20% VRAM
›
S
EXAONE-4.0-1.2B1.3B
Exaone·64K ctx· chat
177
tok/s · 21% VRAM
›
S
OPT 1.3B1.3B
Opt·2K ctx· chat
177
tok/s · 21% VRAM
›
S
MiniCPM-V 4.61.3B
Other·256K ctx· chat· vision
177
tok/s · 21% VRAM
›
S
DeepSeek Coder 1.3B1.35B
Deepseek·16K ctx· coding
171
tok/s · 22% VRAM
›
S
Phi-1 1.3B1.42B
Phi·2K ctx· coding
162
tok/s · 23% VRAM
›
S
Phi-1.5 1.3B1.42B
Phi·2K ctx· chat· coding
162
tok/s · 23% VRAM
›
Next step

Get personalized recommendations.

See ranked models with benchmark scores, run commands, and precise speed estimates for your GeForce GTX 1660 Ti Max-Q.

What this card is worth

GTX 1660 Ti Max-Q holds 139 of the models in our catalogue and is, in practice, a Q4_K_M card — the largest it takes is Falcon-H1 7B at Q4_K_M.

Tokens/sec per $100
—
8B at Q4_K_M, so cards compare like for like
VRAM per $100
—
what memory costs on this card
The biggest it takes
Falcon-H1 7B7.59B · Q4_K_M5.4 GB34 tok/sest
Falcon-H1R 7B7.59B · Q4_K_M5.4 GB34 tok/sest
Falcon Mamba 7B7.27B · Q4_K_M5.0 GB36 tok/sest
WizardLM 2 7B7B · Q4_K_M5.3 GB37 tok/sest
StarCoder2 7B7B · Q4_K_M5.0 GB37 tok/sest
Dolly v2 7B6.9B · Q4_K_M5.3 GB38 tok/sest
granite-4.0-h-tiny 6.9B6.9B · Q4_K_M4.8 GB173 tok/sest
ChatGLM2 6B6.24B · Q5_K_M5.1 GB36 tok/sest
Against cards of similar memory
RTX 3080 10GB10 GB$42917.7 tok/s per $100
Arc B57010 GB$2198.2 tok/s per $100
Radeon RX 670010 GB$29912.0 tok/s per $100
Radeon RX 6750 GRE 10 GB10 GB$22915.7 tok/s per $100
GTX 1660 Ti Max-Q6 GB——

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

NVIDIA GeForce GTX 1660 Ti Max-Q — 6 GB VRAM.

GeForce GTX 1660 Ti Max-Q spec
Brand
NVIDIA
VRAM
6 GB GDDR6
Bandwidth
288 GB/s
FP16 Compute
8.2 TFLOPS
FP32 Compute
4.1 TFLOPS
CUDA Cores
1,536
TDP
60 W
Architecture
Turing
AI capability
139/ 498 models @ Q4

With 6 GB VRAM and 288 GB/s bandwidth, this GPU handles models up to 7B parameters.

Speed ≈ bandwidth / model_size × efficiency. A 7B model at Q4 runs at ~33 tok/s.

Rent it — live prices
Checking what the clouds are charging…
Top models for GeForce GTX 1660 Ti Max-Q
139 fit · showing 20
ModelSizeVRAM Q4tok/sAvg
Alpaca 7B7B4.8 GB3327.7
Baichuan2 7B7B4.8 GB3321.5
Vicuna 7B7B4.8 GB3322.0
MPT-7B7B4.8 GB337.8
Orca 2 7B7B4.8 GB3326.1
WizardLM 2 7B7B4.8 GB3326.1
StarCoder2 7B7B4.8 GB3317.0
WizardCoder Python 7B7B4.8 GB3353.7
WizardLM 7B7B4.8 GB3315.5
OLMo 3.1 RLZero 7B Code7B4.8 GB3321.8
OLMo 3.1 RLZero 7B Math7B4.8 GB3321.8
Dolly v2 7B6.9B4.7 GB337.0
granite-4.0-h-tiny 6.9B6.9B4.7 GB15449.2
Llama 2 7B6.74B4.6 GB3421.1
CodeLlama 7B6.74B4.6 GB3428.1
LLaMA 1 7B6.74B4.6 GB3430.8
DeepSeek Coder 6.7B6.7B4.6 GB3423.6
OPT 6.7B6.7B4.6 GB3418.5
ChatGLM2 6B6.24B4.3 GB3720.7
ChatGLM3 6B6.24B4.3 GB3742.7