NVIDIA GeForce GTX 1660 Ti
No editorial image yet — generic vendor mark shown. Credentials in spec table below.
Turing mid-tier without RT/Tensor cores. 6 GB VRAM fits 7B Q4 with short context. Bandwidth (288 GB/s) is solid for the tier — ~30-40 tok/s on 7B Q4. Same VRAM ceiling as the 1660 Super; the Ti pays for slightly more compute that doesn't help much for inference.
Affiliate disclosure: as an Amazon Associate and partner of other retailers, we earn from qualifying purchases. The verdict on this page is our editorial opinion; affiliate links never influence what we recommend.
Sub-scores sum to 353 / 1000. Headline = 353 × 0.70 (Estimated-confidence discount) = 247. This is an algorithmic performance-tier score — distinct from, and often lower than, the editorial “Our verdict” below, which weighs value and real-world fit (especially for hardware we haven’t measured yet). How scoring works →
Extrapolated from 288 GB/s bandwidth — 34.6 tok/s estimated. No measured benchmarks yet.
Plain-English: Edge-of-fit for 7B; expect compromises.
Verdicts extrapolated from catalog VRAM + bandwidth + ecosystem flags. Hover any chip for the rationale. Want measured numbers? Submit your own run with runlocalai-bench --submit.
This card is for the budget operator who needs a functional local inference rig at the lowest possible entry cost and is willing to accept strict model size limits. The 6 GB VRAM fits a 7B Q4 model with a short context window (2-4K tokens), and the 288 GB/s bandwidth delivers 30-40 tok/s on that workload — usable for chat or code completion. Larger models like 13B Q4 or 7B Q8 are out of reach; the card cannot load them at all. The lack of Tensor cores means no acceleration for CUDA-based inference engines like llama.cpp, but the card still runs them fine via FP16 compute. Pass on this card if you need to run 13B models, want longer context (8K+), or plan to experiment with larger quantizations. At ~$160 used, it is a stopgap for learning local AI, not a long-term investment.
›Why this rating
The GTX 1660 Ti offers decent inference speed for 7B Q4 models at a low price, but its 6 GB VRAM is a hard ceiling that excludes most modern workloads. It scores a 5.5 because it is functional for entry-level use but lacks headroom for growth.
Overview
Turing mid-tier without RT/Tensor cores. 6 GB VRAM fits 7B Q4 with short context. Bandwidth (288 GB/s) is solid for the tier — ~30-40 tok/s on 7B Q4. Same VRAM ceiling as the 1660 Super; the Ti pays for slightly more compute that doesn't help much for inference.
Some links above are affiliate links. We may earn a commission at no extra cost to you. How we make money.
Specs
| VRAM | 6 GB |
| Power draw (peak) | 120 W |
| Released | 2019 |
| MSRP | $279 |
| Backends | CUDA Vulkan |
Models that fit
Open-weight models small enough to run on NVIDIA GeForce GTX 1660 Ti with usable context.
Frequently asked
What models can NVIDIA GeForce GTX 1660 Ti run?
Does NVIDIA GeForce GTX 1660 Ti support CUDA?
How much does NVIDIA GeForce GTX 1660 Ti cost?
Where next?
Reviewed by RunLocalAI Editorial. See our editorial policy for how we research and verify hardware specifications.