A Used RTX 3090 for Local LLMs: What 24GB Buys

Gigabyte GeForce RTX 3090 graphics card, front view

A used RTX 3090 is still the cheapest common way to get 24 GB of VRAM on one NVIDIA card. NVIDIA lists it with 24 GB of GDDR6X on a 384-bit bus, the same capacity as the RTX 4090. For local language models, capacity decides which models fit, so the older card opens the same model sizes as the newer one.

What fits in 24 GB

The llmhardware.io formula is parameters in billions × bytes per parameter × 1.2, plus 2 GB for the system. Run common sizes through it:

  • A 32B model at Q4 comes to about 21.2 GB. It fits.
  • A 27B model at Q4 comes to about 18.2 GB. It fits with room for a longer context.
  • A 14B model at 8-bit comes to about 18.8 GB. It fits.
  • An 8B model at full 16-bit precision comes to about 21.2 GB. It fits.
  • A 70B model at Q4 comes to about 44 GB. It does not.

That 32B line is the reason people still hunt for this card. A 16 GB card misses the 21.2 GB estimate, so the jump from 16 GB to 24 GB is the jump from 14B-class models to 32B-class ones.

Where the newer cards differ

The RTX 4090 has the same 24 GB but a much larger chip: NVIDIA lists 16,384 CUDA cores against 10,496 on the 3090, so you get more compute for the same memory. The RTX 5090 is the first GeForce card in this group to move the capacity line, with 32 GB of GDDR7. That extra 8 GB is room for a longer context on a 32B model, or a 32B model at a higher-quality quant.

Power and fit in your case

NVIDIA lists 350 W of graphics card power for the RTX 3090 and a 750 W required system power. Plan your power supply around that figure, not the one on a seller's listing. If you are thinking about two cards to reach 48 GB, which is what a 70B model at Q4 needs once the 44 GB estimate is split across both, the power, cooling and case space all double too.

What to check on a used card

  • Ask the seller for a photo of the card running a long GPU load with temperatures visible.
  • Check the fans spin up and the card has no bent fins, missing screws or signs it has been opened.
  • Confirm whether any warranty transfers. Many used cards come with none.
  • Once it is in your machine, load a model you know fits and run ollama ps. You want 100% GPU in the PROCESSOR column.

If you want 24 GB and do not need the 4090's extra compute, the 3090 gives you the same model sizes. If you want room past 24 GB on one card, the 5090's 32 GB is the step up.

The RTX 3090 guide has the full spec table and the model list for this card.

Sources: NVIDIA GeForce RTX 3090, RTX 4090 and RTX 5090 spec pages. Memory figures are formula estimates, not benchmarks.

Comments

Post a Comment

Popular posts from this blog

AMD RX 9070 XT for Local LLMs: ROCm Performance Guide (2026)

The AMD RX 9070 XT launched in early 2026 at $599 with 16GB GDDR6 and 648 GB/s memory bandwidth. For local LLM users who want a mid-range GPU without NVIDIA prices, this changes the calculus significantly. Here is what it actually delivers. RX 9070 XT Specs vs Competition GPU VRAM Bandwidth Price LLM Score RX 9070 XT 16 GB 648 GB/s $599 ~60 tok/s RTX 4070 Ti Super 16 GB 672 GB/s $799 ~62 tok/s RTX 4080 Super 16 GB 736 GB/s $999 ~68 tok/s RTX 4090 24 GB 1,008 GB/s ~$1,100 ~90 tok/s At $599 with 16GB VRAM, the RX 9070 XT is the best value GPU for local LLMs in the mid-range tier in 2026. See the full comparison at LLMHardware.io's AMD GPU guide . ROCm Setup on Linux (2026) # Ubuntu 22.04/24.04 sudo apt install rocm-hip-libraries hip-base # or use the AMD installer: wget https://repo.radeon.com/amdgpu-install/6.3/ubuntu/jammy/amdgpu-install_6.3.60300-1_all.deb sudo dpkg -i amdgpu-install*.deb sudo amdgpu-install --usecase=rocm,lrt,openclsdk # Verify ROCm sees your GPU r...
Read more

llama.cpp flash attention: how to enable and benchmark

Image
Flash attention computes attention with far less memory traffic. The math is identical; the memory movement is not. In llama.cpp it matters most in two situations: long context windows and tight video memory (VRAM). With flash attention enabled, the key-value cache shrinks substantially. A 32k-token context window becomes realistic on a 16GB card, and the prompt-processing phase speeds up, so long documents get ingested faster. The catch is compatibility: not every model and quantization combination supports it, and on some setups the speedup is modest. If you run short chats on a strong GPU, you may barely notice. Where it shines is retrieval-augmented generation (RAG) and long-document work on mid-range cards. Enable it, compare your tokens per second, and keep it if your workload benefits. For the common case of pasting a long PDF and asking questions, it is close to free performance. See how context length changes VRAM needs .
Read more

SillyTavern with Local LLMs: Complete Setup for AI Companions

Image
SillyTavern is the most feature-rich frontend for character-based AI conversations. What began as an AI roleplay tool has evolved into a powerful platform with persona management, memory systems, extensive prompt engineering tools, and connections to every major local inference backend. Here's how to set it up with local models. Quick connect: SillyTavern connects to Ollama, KoboldCpp, llama.cpp server, and TabbyAPI (ExLlamaV2) out of the box. Any of the hardware setups at LLMHardware.io provides a solid backend for SillyTavern. Installation Install via git: git clone https://github.com/SillyTavern/SillyTavern && cd SillyTavern && npm install then node server.js . Access at localhost:8000. For persistent installation, use the included start.sh script. SillyTavern uses Node.js and runs on any platform. The companion app (STMP) enables mobile access to your SillyTavern instance via Tailscale or local network. Character Cards and the TavernAI Format Chara...
Read more