Best GPU for LTX-2.5 (22B)
Which graphics cards run LTX-2.5 (22B) well, which run it at all, and the cheapest of each by launch price.
Cheapest NVIDIA card that runs it wellRTX 5090 32 GBRuns well Q8_0Launch price $1,999 · 32 GBCheapest NVIDIA card that runs it at allRTX 5060 Ti 16 GBTight Q2_KLaunch price $429 · 16 GB
Cheapest AMD or Intel card that runs it wellNoneNo consumer card in the list does this.
Ranked by memory verdict, then by the maker's original launch price in USD — only to compare cards with each other; today's street and used prices differ. NVIDIA is listed first because most ComfyUI workflows and custom nodes are built and tested on it; AMD and Intel run ComfyUI officially, with more setup (details). The verdicts are about memory, not speed.
01Every desktop card, best first
| GPU | VRAM | Verdict | Best file | Launch price |
|---|---|---|---|---|
| Arc Pro B70 32 GB | 32 GB | Runs well | Q8_0 | $949 |
| RTX 5090 32 GB | 32 GB | Runs well | Q8_0 | $1,999 |
| RX 7900 XT 20 GB | 20 GB | Runs | Q4_K_M | $899 |
| RX 7900 XTX 24 GB | 24 GB | Runs | Q6_K | $999 |
| RTX 3090 24 GB | 24 GB | Runs | Q6_K | $1,499 |
| RTX 4090 24 GB | 24 GB | Runs | Q6_K | $1,599 |
| RTX 3090 Ti 24 GB | 24 GB | Runs | Q6_K | $1,999 |
| Arc Pro B60 24 GB | 24 GB | Runs | Q6_K | — |
| RX 7600 XT 16 GB | 16 GB | Tight | Q2_K | $329 |
| RX 9060 XT 16 GB | 16 GB | Tight | Q2_K | $349 |
| Arc A770 16 GB | 16 GB | Tight | Q2_K | $349 |
| RTX 5060 Ti 16 GB | 16 GB | Tight | Q2_K | $429 |
| RTX 4060 Ti 16 GB | 16 GB | Tight | Q2_K | $499 |
| RX 7800 XT 16 GB | 16 GB | Tight | Q2_K | $499 |
| RX 9070 16 GB | 16 GB | Tight | Q2_K | $549 |
| RX 7900 GRE 16 GB | 16 GB | Tight | Q2_K | $549 |
| RX 9070 XT 16 GB | 16 GB | Tight | Q2_K | $599 |
| RTX 5070 Ti 16 GB | 16 GB | Tight | Q2_K | $749 |
| RTX 4070 Ti Super 16 GB | 16 GB | Tight | Q2_K | $799 |
| RTX 5080 16 GB | 16 GB | Tight | Q2_K | $999 |
| RTX 4080 Super 16 GB | 16 GB | Tight | Q2_K | $999 |
| RTX 4080 16 GB | 16 GB | Tight | Q2_K | $1,199 |
| Arc B570 10 GB | 10 GB | Offload only | Q4_K_M | $219 |
| RTX 5050 8 GB | 8 GB | Offload only | Q4_K_M | $249 |
| RTX 3050 8 GB | 8 GB | Offload only | Q4_K_M | $249 |
| Arc B580 12 GB | 12 GB | Offload only | Q2_K | $249 |
| RX 7600 8 GB | 8 GB | Offload only | Q4_K_M | $269 |
| RX 9050 8 GB | 8 GB | Offload only | Q4_K_M | $279 |
| RTX 5060 8 GB | 8 GB | Offload only | Q4_K_M | $299 |
| RTX 4060 8 GB | 8 GB | Offload only | Q4_K_M | $299 |
| RX 9060 XT 8 GB | 8 GB | Offload only | Q4_K_M | $299 |
| RTX 3060 12 GB | 12 GB | Offload only | Q2_K | $329 |
| RTX 5060 Ti 8 GB | 8 GB | Offload only | Q4_K_M | $379 |
| RTX 4060 Ti 8 GB | 8 GB | Offload only | Q4_K_M | $399 |
| RTX 3060 Ti 8 GB | 8 GB | Offload only | Q4_K_M | $399 |
| RX 7700 XT 12 GB | 12 GB | Offload only | Q2_K | $449 |
| RTX 3070 8 GB | 8 GB | Offload only | Q4_K_M | $499 |
| RTX 2070 Super 8 GB | 8 GB | Offload only | Q4_K_M | $499 |
| RTX 5070 12 GB | 12 GB | Offload only | Q2_K | $549 |
| RX 9070 GRE 12 GB | 12 GB | Offload only | Q2_K | $549 |
| RTX 4070 Super 12 GB | 12 GB | Offload only | Q2_K | $599 |
| RTX 4070 12 GB | 12 GB | Offload only | Q2_K | $599 |
| RTX 3070 Ti 8 GB | 8 GB | Offload only | Q4_K_M | $599 |
| RTX 3080 10 GB | 10 GB | Offload only | Q4_K_M | $699 |
| RTX 2080 Super 8 GB | 8 GB | Offload only | Q4_K_M | $699 |
| RTX 4070 Ti 12 GB | 12 GB | Offload only | Q2_K | $799 |
| RTX 2080 Ti 11 GB | 11 GB | Offload only | Q4_K_M | $999 |
| RTX 3080 Ti 12 GB | 12 GB | Offload only | Q2_K | $1,199 |
| RTX 3080 12 GB | 12 GB | Offload only | Q2_K | — |
| RTX 3060 8 GB | 8 GB | Offload only | Q4_K_M | — |
| RTX 2060 12 GB | 12 GB | Offload only | Q2_K | — |
| RTX 2060 6 GB | 6 GB | Not practical | Q2_K | $349 |
| RTX 3050 6 GB | 6 GB | Not practical | Q2_K | — |
02Before you buy
For local AI, memory decides what you can load; speed decides how long you wait. A slower card with more VRAM runs models a faster card cannot load at all. Also check system RAM: LTX-2.5 with its text encoder wants about 48 GB of RAM at the file a 16 GB card would use. All LTX-2.5 files →