// RUN ON CONSUMER GPU

Run Phi-4 on Consumer GPU hardware

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

HardwareEst tok/sFits 8k?Max ctx~Price
NVIDIA GeForce RTX 5090 → ~198.9 yes 123k ~$1,999
NVIDIA GeForce RTX 4090 → ~111.9 yes 78k ~$1,599
NVIDIA GeForce RTX 3090 Ti → ~111.9 yes 78k ~$1,999
NVIDIA GeForce RTX 5080 → ~106.5 yes 33k ~$999
NVIDIA GeForce RTX 3090 → ~103.9 yes 78k ~$1,499
NVIDIA GeForce RTX 3080 Ti → ~101.2 yes 10k ~$1,199
NVIDIA GeForce RTX 5070 Ti → ~99.4 yes 33k ~$749
AMD Radeon RX 7900 XTX → ~94.3 yes 78k ~$999
NVIDIA GeForce RTX 4080 → ~81.7 yes 33k ~$1,199
AMD Radeon RX 7900 XT → ~78.6 yes 55k ~$899
NVIDIA GeForce RTX 5070 → ~74.6 yes 10k ~$549
NVIDIA GeForce RTX 4070 Ti Super → ~74.6 yes 33k ~$799
AMD Radeon RX 9070 XT → ~63.3 yes 33k ~$599
AMD Radeon RX 7800 XT → ~61.3 yes 33k ~$499
NVIDIA GeForce RTX 4070 Super → ~55.9 yes 10k ~$599
NVIDIA GeForce RTX 4070 → ~55.9 yes 10k ~$599
Intel Arc Pro B70 → ~55.8 yes 123k —
Intel Arc A770 16GB → ~51.4 yes 33k ~$329
AMD Radeon RX 6800 XT → ~50.3 yes 33k ~$649
AMD Radeon RX 6900 XT → ~50.3 yes 33k ~$999
NVIDIA GeForce RTX 5060 Ti 16GB → ~49.7 yes 33k ~$429
Intel Arc Pro B60 → ~41.9 yes 78k —
Intel Arc B580 12GB → ~41.9 yes 10k ~$249
NVIDIA GeForce RTX 4060 Ti 16GB → ~32 yes 33k ~$499
NVIDIA DGX Spark → ~30.3 yes 646k ~$3,999
AMD Ryzen AI Max+ 395 → ~25.1 yes 646k ~$1,999
Intel Arc Pro B50 → ~20.6 yes 33k —

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

→ Phi-4 overview → All hardware

FAQ

What Consumer GPU hardware runs Phi-4?▶

On GenAIList, 27 Consumer GPU options can run Phi-4 at Q4 — the fastest is the NVIDIA GeForce RTX 5090 at ~198.9 tok/s.

How fast is Phi-4 on Consumer GPU hardware?▶

Estimated generation speed ranges from ~20.6 to ~198.9 tok/s at Q4 (memory-bandwidth bound). Click any option for its full estimate and any measured numbers.