Guide
Used RTX 3090 buying guide (EU)
A used RTX 3090 is the 24 GB card this catalog still treats as the sensible single GPU, and as one half of a 70B rig. EUR amounts include VAT and are not live shop listings. USD is an FX estimate.
Three rows, all unsourced prices
- GeForce RTX 3090 24GB: $806 estimate, €720 incl. VAT. Global prices (USD), EUR available. Dated 2026-10-14. Marked used. Global prices (USD), EUR available. EUR includes VAT. USD is an FX estimate. No shop listing was checked. sourceUrl is null.
- Used PC with RTX 3090 24GB: $1,433 estimate, €1,280 incl. VAT. Global prices (USD), EUR available. Dated 2026-10-14. Marked used. Global prices (USD), EUR available. EUR includes VAT. USD is an FX estimate. No shop listing was checked. sourceUrl is null.
- Dual RTX 3090, 48 GB: $2,217 estimate, €1,980 incl. VAT. Global prices (USD), EUR available. Dated 2026-10-14. Marked used. Global prices (USD), EUR available. EUR includes VAT. USD is an FX estimate. No shop listing was checked. sourceUrl is null.
GeForce RTX 3090 24GB, used; Ryzen 7 5700X or similar; 64 GB DDR4; 850W PSU, decent airflow. 2× GeForce RTX 3090 24GB, used; Ryzen 9 5900X or similar; 128 GB DDR4; 1200W PSU, a case that can breathe. Those part lists are the builds in the catalog, not a teardown of a listing. TDP on the single card row is 350 W with no separate source URL.
GeForce RTX 3090 24GB lists 24 GB and 936 GB/s. bandwidthEstimated is false, but bandwidthSourceUrl is null, so that speed is not a citation.
What 24 GB holds at Q4
Used PC with RTX 3090 24GB has 22.8 GB usable after the reserve in this formula (catalog memory 24 GB, vram). Qwen2.5 32B Q4 at 8,192 tokens is 21.3 GB in this formula (18.4 GB weights, 2.1 GB KV cache, 0.8 GB overhead).
Full fit on the used 3090 PC at Q4 and 8,192 tokens: Llama 3.2 3B, Llama 3.1 8B, Mistral 7B v0.3, Gemma 2 9B, Qwen2.5 14B, Phi-4 14B, Gemma 2 27B, Qwen2.5 32B.
Offload only on that PC, which is not a full fit: Llama 3.3 70B, Qwen2.5 72B, DeepSeek-R1-Distill 70B. The calculator does not prefer offload while a full fit exists.
When the second card is the actual product
Llama 3.3 70B Q4 at 8,192 tokens is 43.0 GB in this formula (39.5 GB weights, 2.7 GB KV cache, 0.8 GB overhead). Fit on one 3090 PC: offload only, not a full fit. Fit on Dual RTX 3090, 48 GB: a full fit (usable 45.6 GB). The catalog price for that pair is $2,217 estimate, €1,980 incl. VAT, still with sourceUrl null. Two used cards also mean two power limits in the catalog (700 W combined) and a case the parts list only describes as able to breathe.
What we did not check
No shop listing, no serial-number check, no failure rate for ex-mining cards, no measured tokens/s. Indicative prices move, and the note on every price says no shop listing was checked. Use the hardware page search links if you want a current shop page. Those links are not affiliate sales unless a real partner id exists, and none does.
Run it: Qwen2.5 32B on the calculator, Llama 70B under €2,000, hardware catalog, affiliate disclosure.
Related: Best GPU for local LLMs in 2026, How much VRAM does a 70B model need?, Mac vs PC for a local LLM.