vram.run Models Hardware Providers Cloud State of Inference
API provider data is live · Hardware & cloud pricing curated 2026-02-23

ko-sroberta-multitask

by jhgan

111M params · sentence-similarity · 149 likes · 963.6k downloads

ko-sroberta-multitask is a 111M parameter model. At Q4 quantization it requires 0GB of VRAM. It runs comfortably on GeForce RTX 4090 (11869 tok/s), GeForce RTX 5090 (17799 tok/s), M4 Max 128GB (4343 tok/s).

Inference providers

Provider$/1M in$/1M outThroughput
HF Inference

GPU compatibility

GPUVRAMQ4 DecodeVerdict
GeForce RTX 409024GB11869 tok/scomfortable
GeForce RTX 509032GB17799 tok/scomfortable
M4 Max 128GB128GB4343 tok/scomfortable
M4 Pro 48GB48GB2171 tok/scomfortable
M4 Pro 24GB24GB2171 tok/scomfortable
A100 PCIe 80 GB80GB21746 tok/scomfortable
H100 SXM5 80 GB80GB43739 tok/scomfortable
GeForce RTX 309024GB10494 tok/scomfortable
Radeon RX 7900 XTX24GB8678 tok/scomfortable
GeForce RTX 408016GB8423 tok/scomfortable
Install CLI [email protected] Raw data · MIT · API data: live · HW/Cloud data: curated 2026-02-23 · v0.6.0