# NVIDIA GeForce RTX 3090

116 Hacker News commenters recommended NVIDIA GeForce RTX 3090 and 21 criticized it, November 2025 to November 2026. It was recommended in 12 of 13 months; owners who said how long report a median of 5 years (14 owners). It meets the Hacker Approved badge rule.

**Verdict:** A used RTX 3090 is the go-to card for running local LLMs, with 24GB of VRAM, strong price-performance, and better speed than Apple hardware, though used prices have climbed.

## Why people recommend it

- Top pick for local LLMs (8 comments): "if you want the best possible AI machine at below $5k... actually you should probably just buy an RTX 3090 or 5090" ([source](https://news.ycombinator.com/item?id=48808849)) "A used 3090 is probably the most recommended card in the local LLM community right now, and for good reason." ([source](https://news.ycombinator.com/item?id=47524956)) "3090 gets 100+ tokens/second for QWEN, very close to what you would see with a cloud based model." ([source](https://news.ycombinator.com/item?id=47395863))
- Great value bought used (6 comments): "I did this with used parts and cheaper consumer cards (3090s) and did much of the same calculations. I found it was way cheaper for me as well." ([source](https://news.ycombinator.com/item?id=48228554)) "a dual 3090 setup which you can build for $3k with parts from facebook marketplace and run in your basement" ([source](https://news.ycombinator.com/item?id=48780933)) "A couple of 3090s at about 800-900 euros each, as much RAM as you can afford is king right now." ([source](https://news.ycombinator.com/item?id=49343348))
- Outperforms Apple hardware (5 comments): "If you compare the performance of a 3090 and above vs any Apple hardware you would be insane to go with the Apple hardware." ([source](https://news.ycombinator.com/item?id=47542603)) "two of them (or three 3090/4090s) have the same total memory as your Mac, are in the same ballpark in price" ([source](https://news.ycombinator.com/item?id=46154264)) "A single 3090 will run Qwen3.6 27b fine, and its VRAM speed is twice of what the best Mac has." ([source](https://news.ycombinator.com/item?id=48725001))
- Scales well with multiple cards (5 comments): "a dual 3090 setup which you can build for $3k with parts from facebook marketplace and run in your basement" ([source](https://news.ycombinator.com/item?id=48780933)) "the 3090 supports NVLink, which is actually more useful for inference speed than native BF16 support" ([source](https://news.ycombinator.com/item?id=46358573)) "I also get a boost in tokens per second (not double, about 1.5x compared to a single GPU)." ([source](https://news.ycombinator.com/item?id=48740665))

## What people warn about

- VRAM limits context and model size (5 comments): "3090 and 2x3090 are quite popular. But if you uses gigantic (for local models) context of 200k it will go south pretty quickly" ([source](https://news.ycombinator.com/item?id=48583192)) "That 3090 is going to burn 750W and it will still cap you at a 4 bit quant and ~48K context." ([source](https://news.ycombinator.com/item?id=48723254)) "48GB of VRAM is just not enough to use a local LLM like a cloud model" ([source](https://news.ycombinator.com/item?id=49985119))
- Used prices are high (3 comments): "price on those is unfortunately high, in fact so high that getting a new Radeon AI PRO R9700 might be a better deal" ([source](https://news.ycombinator.com/item?id=49272523)) "even the most often recommended GPU which is a 3090 with 24 GB VRAM, you can find used going for $700 at minimum" ([source](https://news.ycombinator.com/item?id=47483314)) "The 3090's in my area seem pretty expensive for a several year old second hand card" ([source](https://news.ycombinator.com/item?id=47836007))
- Linux driver problems (3 comments): "do all nvidia cards have the same issue with Linux drivers, where the fan won't ever go below 30%?" ([source](https://news.ycombinator.com/item?id=46238518)) "AFAIK this comes down a lot to NVIDIA not doing enough efforts for the Linux drivers" ([source](https://news.ycombinator.com/item?id=46463436)) "I had this problem on Fedora and my RTX 3090. Never got to the bottom of it." ([source](https://news.ycombinator.com/item?id=49403937))
- Power hungry (2 comments): "That 3090 is going to burn 750W and it will still cap you at a 4 bit quant and ~48K context." ([source](https://news.ycombinator.com/item?id=48723254)) "It's a little bit power hungry but fairly impressive performance wise." ([source](https://news.ycombinator.com/item?id=46253628))

## Alternatives hackers recommend

- [NVIDIA GeForce RTX 5090](https://hackerapproved.com/p/nvidia-geforce-rtx-5090.md): 102 recommend
- [NVIDIA RTX PRO 6000](https://hackerapproved.com/p/nvidia-rtx-pro-6000.md): 77 recommend
- [Nvidia GPUs](https://hackerapproved.com/p/nvidia-gpus.md): 67 recommend
- [AMD Radeon GPU](https://hackerapproved.com/p/amd-gpus.md): 48 recommend
- [AMD Radeon AI Pro R9700](https://hackerapproved.com/p/amd-radeon-ai-pro-r9700.md): 45 recommend

## Where to get it

- [See price on Amazon](https://www.amazon.com/s?k=NVIDIA+GeForce+RTX+3090+graphics+card&tag=hackerapprove-20) (affiliate link)
- [Official site](https://www.nvidia.com/en-us/geforce/graphics-cards/30-series/rtx-3090-3090ti/)

## Cite as

Hacker Approved, "NVIDIA GeForce RTX 3090", https://hackerapproved.com/p/nvidia-geforce-rtx-3090/ (counts from 1,795,640 Hacker News comments, November 2025 to November 2026).