// RUN ON CONSUMER GPU

Run CogVLM-17B on Consumer GPU hardware

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

HardwareEst tok/sFits 8k?Max ctx~Price
NVIDIA GeForce RTX 5090 → ~163.8 yes 70k ~$1,999
NVIDIA GeForce RTX 4090 → ~92.1 yes 42k ~$1,599
NVIDIA GeForce RTX 3090 Ti → ~92.1 yes 42k ~$1,999
NVIDIA GeForce RTX 5080 → ~87.7 yes 14k ~$999
NVIDIA GeForce RTX 3090 → ~85.5 yes 42k ~$1,499
NVIDIA GeForce RTX 5070 Ti → ~81.9 yes 14k ~$749
AMD Radeon RX 7900 XTX → ~77.6 yes 42k ~$999
NVIDIA GeForce RTX 4080 → ~67.3 yes 14k ~$1,199
AMD Radeon RX 7900 XT → ~64.7 yes 28k ~$899
NVIDIA GeForce RTX 4070 Ti Super → ~61.4 yes 14k ~$799
AMD Radeon RX 9070 XT → ~52.1 yes 14k ~$599
AMD Radeon RX 7800 XT → ~50.5 yes 14k ~$499
Intel Arc Pro B70 → ~46 yes 70k —
Intel Arc A770 16GB → ~42.4 yes 14k ~$329
AMD Radeon RX 6800 XT → ~41.4 yes 14k ~$649
AMD Radeon RX 6900 XT → ~41.4 yes 14k ~$999
NVIDIA GeForce RTX 5060 Ti 16GB → ~40.9 yes 14k ~$429
Intel Arc Pro B60 → ~34.5 yes 42k —
NVIDIA GeForce RTX 4060 Ti 16GB → ~26.3 yes 14k ~$499
NVIDIA DGX Spark → ~24.9 yes 397k ~$3,999
AMD Ryzen AI Max+ 395 → ~20.7 yes 397k ~$1,999
Intel Arc Pro B50 → ~16.9 yes 14k —

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

→ CogVLM-17B overview → All hardware

FAQ

What Consumer GPU hardware runs CogVLM-17B?▶

On GenAIList, 22 Consumer GPU options can run CogVLM-17B at Q4 — the fastest is the NVIDIA GeForce RTX 5090 at ~163.8 tok/s.

How fast is CogVLM-17B on Consumer GPU hardware?▶

Estimated generation speed ranges from ~16.9 to ~163.8 tok/s at Q4 (memory-bandwidth bound). Click any option for its full estimate and any measured numbers.