Best used graphics cards for local AI

Ranked by what each one runs fully in memory at 32k context, with prices in Germany today.

In short

The top pick in Germany right now is the NVIDIA GeForce RTX 5090, 5.599 € at eBay (used, asking price): it runs Qwen 3.8 27B fully in memory at 32k context at about 35 tok/s (estimated). Ranked by the most capable open model each one runs fully in memory at 32k context and a usable speed, then by how many models it runs, then by speed on Qwen 3.8 27B, then by price.

Updated . Speeds are estimates; how they are worked out.

The list

Best first. Used prices are eBay asking prices. Speeds are estimates est.

#CardMemoryPriceBest model it runs at 32kModels in memoryQwen 3.8 27Btok/s per 1.000 €
1NVIDIA GeForce RTX 509032 GB5.599 €eBay, askingQwen 3.8 27BQ8_035 tok/s2935 tok/s6.3
2NVIDIA GeForce RTX 3090 Ti24 GB1.599 €eBay, askingQwen 3.8 27BUD-Q5_K_M28 tok/s2628 tok/s17.7
3NVIDIA GeForce RTX 409024 GB2.800 €eBay, askingQwen 3.8 27BUD-Q5_K_M28 tok/s2628 tok/s10.1
4NVIDIA GeForce RTX 309024 GB1.290 €eBay, askingQwen 3.8 27BUD-Q5_K_M26 tok/s2626 tok/s20.4
5AMD Radeon RX 7900 XTX24 GB900 €eBay, askingQwen 3.8 27BUD-Q5_K_M24 tok/s2624 tok/s26.6
6AMD Radeon RX 7900 XT20 GB725 €eBay, askingQwen 3.8 27BUD-Q4_K_M24 tok/s1824 tok/s32.4
7AMD Radeon AI PRO R970032 GB2.225 €eBay, askingQwen 3.6 35B-A3BUD-Q6_K66 tok/s2011 tok/s5
8NVIDIA Tesla P4024 GB369 €eBay, askingQwen 3.6 35B-A3BUD-Q4_K_M50 tok/s1510 tok/s26.3
9Intel Arc A770 16GB16 GB350 €eBay, askinggpt-oss 20BMXFP448 tok/s13offload–
10AMD Radeon RX 7800 XT16 GB600 €eBay, askinggpt-oss 20BMXFP464 tok/s13offload–
11NVIDIA GeForce RTX 5060 Ti 16GB16 GB660 €eBay, askinggpt-oss 20BMXFP452 tok/s13offload–
12AMD Radeon RX 9070 XT16 GB800 €eBay, askinggpt-oss 20BMXFP465 tok/s13offload–

Questions

What is the best used graphics card for local AI in Germany?
The top pick in Germany right now is the NVIDIA GeForce RTX 5090, 5.599 € at eBay (used, asking price): it runs Qwen 3.8 27B fully in memory at 32k context at about 35 tok/s (estimated).
Which gives the most speed for the money?
The AMD Radeon RX 7900 XT: about 24 tok/s on Qwen 3.8 27B for 725 € (used, asking price), the most tokens per second per unit of price on this list (estimated).
What is the cheapest option here that runs Qwen 3.8 27B fully in memory?
The NVIDIA Tesla P40, 369 € at eBay (used, asking price), at about 10 tok/s at 32k context (estimated).

Other lists