NVIDIA GeForce RTX 3050
No editorial image yet — generic vendor mark shown. Credentials in spec table below.
Ampere entry with full Tensor + RT cores at the $200 tier. 8 GB VRAM is the practical floor for serious 7B work. Bandwidth (224 GB/s) is the bottleneck — ~40-55 tok/s on 7B Q4. Often shipped in pre-built office PCs; the 'I have it because it came with the system' audience.
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 375 / 1000. Headline = 375 × 0.70 (Estimated-confidence discount) = 263. 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 224 GB/s bandwidth — 26.9 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 operator who already has it in a pre-built or budget build and wants to run small local models without spending more. It handles 7B Q4 models at 40-55 tok/s, usable for chat and light code generation. 8 GB VRAM is the floor for 7B; 13B models only fit at Q3 or lower, with degraded quality. The 224 GB/s bandwidth limits throughput—don't expect fast responses on larger models. Pass if you're buying new for local AI; a used RTX 3060 12GB or RX 6700 XT offers more VRAM and bandwidth for similar money. At $200 used, it's acceptable only if it's already in the rig.
›Why this rating
The RTX 3050 is severely bandwidth-limited and VRAM-constrained, making it a poor choice for serious local AI work. It barely handles 7B models and fails on 13B+ unless heavily quantized. The rating reflects its viability only as a free or incidental card.
Overview
Ampere entry with full Tensor + RT cores at the $200 tier. 8 GB VRAM is the practical floor for serious 7B work. Bandwidth (224 GB/s) is the bottleneck — ~40-55 tok/s on 7B Q4. Often shipped in pre-built office PCs; the 'I have it because it came with the system' audience.
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) | 130 W |
| Released | 2022 |
| MSRP | $249 |
| Backends | CUDA Vulkan |
Models that fit
Open-weight models small enough to run on NVIDIA GeForce RTX 3050 with usable context.
Frequently asked
What models can NVIDIA GeForce RTX 3050 run?
Does NVIDIA GeForce RTX 3050 support CUDA?
How much does NVIDIA GeForce RTX 3050 cost?
Where next?
Reviewed by RunLocalAI Editorial. See our editorial policy for how we research and verify hardware specifications.