NVIDIA GeForce RTX 2070
Turing high-tier. 8 GB VRAM, similar bandwidth to the 2060 Super, slightly more compute. Runs 7B Q4 at ~65-80 tok/s, 13B Q4 with light offload. A solid used-market choice for operators with $200-260 to spend.
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 461 / 1000. Headline = 461 × 0.70 (Estimated-confidence discount) = 323. 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 448 GB/s bandwidth — 53.8 tok/s estimated. No measured benchmarks yet.
Plain-English: Comfortable for 7B chat.
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 operator who needs a reliable, affordable entry into local inference with CUDA-only stacks, and is willing to work within an 8 GB VRAM ceiling. It handles 7B Q4 models at 50-65 tok/s and 13B Q4 with partial offload at 20-30 tok/s, making it a solid choice for chat and code completion on smaller models. The 8 GB VRAM breaks on 30B+ models and struggles with 13B Q4 fully in VRAM, requiring offloading that slows generation. Pass if you need to run 13B models entirely on GPU or plan to scale to 30B+; the 3060 12 GB offers more VRAM for similar money. At ~$240 used, this is a fair value for a CUDA-only rig that prioritizes speed on small models over capacity.
›Why this rating
The RTX 2070 offers good speed for 7B models and a low used price, but its 8 GB VRAM limits model size and forces offloading for 13B. It's a competent entry-level card, not a standout for larger workloads.
Overview
Turing high-tier. 8 GB VRAM, similar bandwidth to the 2060 Super, slightly more compute. Runs 7B Q4 at ~65-80 tok/s, 13B Q4 with light offload. A solid used-market choice for operators with $200-260 to spend.
Some links above are affiliate links. We may earn a commission at no extra cost to you. How we make money.
Specs
| VRAM | 8 GB |
| Power draw (peak) | 175 W |
| Released | 2018 |
| MSRP | $499 |
| Backends | CUDA Vulkan |
Models that fit
Open-weight models small enough to run on NVIDIA GeForce RTX 2070 with usable context.
Frequently asked
What models can NVIDIA GeForce RTX 2070 run?
Does NVIDIA GeForce RTX 2070 support CUDA?
How much does NVIDIA GeForce RTX 2070 cost?
Where next?
Reviewed by RunLocalAI Editorial. See our editorial policy for how we research and verify hardware specifications.