// RUN ON CONSUMER GPU

Run RoBERTa Base on Consumer GPU hardware

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

HardwareEst tok/sFits 8k?Max ctx~Price
NVIDIA GeForce RTX 5090 → ~22,272 yes 425k ~$1,999
NVIDIA GeForce RTX 4090 → ~12,528 yes 313k ~$1,599
NVIDIA GeForce RTX 3090 Ti → ~12,528 yes 313k ~$1,999
NVIDIA GeForce RTX 5080 → ~11,931.4 yes 201k ~$999
NVIDIA GeForce RTX 3090 → ~11,633.1 yes 313k ~$1,499
NVIDIA GeForce RTX 3080 Ti → ~11,334.9 yes 145k ~$1,199
NVIDIA GeForce RTX 5070 Ti → ~11,136 yes 201k ~$749
AMD Radeon RX 7900 XTX → ~10,560 yes 313k ~$999
NVIDIA GeForce RTX 3080 10GB → ~9,445.7 yes 116k ~$699
NVIDIA GeForce RTX 4080 → ~9,147.4 yes 201k ~$1,199
AMD Radeon RX 7900 XT → ~8,800 yes 257k ~$899
NVIDIA GeForce RTX 5070 → ~8,352 yes 145k ~$549
NVIDIA GeForce RTX 4070 Ti Super → ~8,352 yes 201k ~$799
NVIDIA GeForce RTX 3070 Ti → ~7,556.6 yes 88k ~$599
AMD Radeon RX 9070 XT → ~7,084 yes 201k ~$599
AMD Radeon RX 7800 XT → ~6,864 yes 201k ~$499
NVIDIA GeForce RTX 4070 Super → ~6,264 yes 145k ~$599
NVIDIA GeForce RTX 4070 → ~6,264 yes 145k ~$599
Intel Arc Pro B70 → ~6,253.7 yes 425k —
Intel Arc A770 16GB → ~5,760 yes 201k ~$329
AMD Radeon RX 6800 XT → ~5,632 yes 201k ~$649
AMD Radeon RX 6900 XT → ~5,632 yes 201k ~$999
NVIDIA GeForce RTX 3070 → ~5,568 yes 88k ~$499
NVIDIA GeForce RTX 5060 Ti 16GB → ~5,568 yes 201k ~$429
NVIDIA GeForce RTX 5060 Ti 8GB → ~5,568 yes 88k ~$379
Intel Arc Pro B60 → ~4,690.3 yes 313k —
Intel Arc B580 12GB → ~4,690.3 yes 145k ~$249
NVIDIA GeForce RTX 4060 Ti 8GB → ~3,579.4 yes 88k ~$399
NVIDIA GeForce RTX 4060 Ti 16GB → ~3,579.4 yes 201k ~$499
NVIDIA DGX Spark → ~3,393 yes 1734k ~$3,999
AMD Ryzen AI Max+ 395 → ~2,816 yes 1734k ~$1,999
Intel Arc Pro B50 → ~2,304 yes 201k —
NVIDIA GeForce RTX 2050 Mobile → ~1,392 yes 32k —

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

→ RoBERTa Base overview → All hardware

FAQ

What Consumer GPU hardware runs RoBERTa Base?▶

On GenAIList, 33 Consumer GPU options can run RoBERTa Base at Q4 — the fastest is the NVIDIA GeForce RTX 5090 at ~22,272 tok/s.

How fast is RoBERTa Base on Consumer GPU hardware?▶

Estimated generation speed ranges from ~1,392 to ~22,272 tok/s at Q4 (memory-bandwidth bound). Click any option for its full estimate and any measured numbers.