// RUN ON CONSUMER GPU

Run LLaVA-CoT on Consumer GPU hardware

28 options ranked by estimated speed at Q4 / 8k context.

HardwareEst tok/sFits 8k?Max ctx~Price
NVIDIA GeForce RTX 5090 → ~253.1 yes 133k ~$1,999
NVIDIA GeForce RTX 4090 → ~142.4 yes 88k ~$1,599
NVIDIA GeForce RTX 3090 Ti → ~142.4 yes 88k ~$1,999
NVIDIA GeForce RTX 5080 → ~135.6 yes 43k ~$999
NVIDIA GeForce RTX 3090 → ~132.2 yes 88k ~$1,499
NVIDIA GeForce RTX 3080 Ti → ~128.8 yes 21k ~$1,199
NVIDIA GeForce RTX 5070 Ti → ~126.5 yes 43k ~$749
AMD Radeon RX 7900 XTX → ~120 yes 88k ~$999
NVIDIA GeForce RTX 3080 10GB → ~107.3 yes 9k ~$699
NVIDIA GeForce RTX 4080 → ~103.9 yes 43k ~$1,199
AMD Radeon RX 7900 XT → ~100 yes 66k ~$899
NVIDIA GeForce RTX 5070 → ~94.9 yes 21k ~$549
NVIDIA GeForce RTX 4070 Ti Super → ~94.9 yes 43k ~$799
AMD Radeon RX 9070 XT → ~80.5 yes 43k ~$599
AMD Radeon RX 7800 XT → ~78 yes 43k ~$499
NVIDIA GeForce RTX 4070 Super → ~71.2 yes 21k ~$599
NVIDIA GeForce RTX 4070 → ~71.2 yes 21k ~$599
Intel Arc Pro B70 → ~71.1 yes 133k —
Intel Arc A770 16GB → ~65.5 yes 43k ~$329
AMD Radeon RX 6800 XT → ~64 yes 43k ~$649
AMD Radeon RX 6900 XT → ~64 yes 43k ~$999
NVIDIA GeForce RTX 5060 Ti 16GB → ~63.3 yes 43k ~$429
Intel Arc Pro B60 → ~53.3 yes 88k —
Intel Arc B580 12GB → ~53.3 yes 21k ~$249
NVIDIA GeForce RTX 4060 Ti 16GB → ~40.7 yes 43k ~$499
NVIDIA DGX Spark → ~38.6 yes 656k ~$3,999
AMD Ryzen AI Max+ 395 → ~32 yes 656k ~$1,999
Intel Arc Pro B50 → ~26.2 yes 43k —

Estimate at Q4 (memory-bound). Click any option for its full per-hardware estimate and measured numbers.

→ LLaVA-CoT overview → All hardware

FAQ

What Consumer GPU hardware runs LLaVA-CoT?▶

On GenAIList, 28 Consumer GPU options can run LLaVA-CoT at Q4 — the fastest is the NVIDIA GeForce RTX 5090 at ~253.1 tok/s.

How fast is LLaVA-CoT on Consumer GPU hardware?▶

Estimated generation speed ranges from ~26.2 to ~253.1 tok/s at Q4 (memory-bandwidth bound). Click any option for its full estimate and any measured numbers.