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

LFM2.5-1.2B-Instruct

by LiquidAI

1.2B params · text-generation · 614 likes · 144.4k downloads

LFM2.5-1.2B-Instruct is a 1.2B parameter model. At Q4 quantization it requires 1GB of VRAM. It runs comfortably on GeForce RTX 4090 (1121 tok/s), GeForce RTX 5090 (1682 tok/s), M4 Max 128GB (410 tok/s).

GPU compatibility

GPUVRAMQ4 DecodeVerdict
GeForce RTX 409024GB1121 tok/scomfortable
GeForce RTX 509032GB1682 tok/scomfortable
M4 Max 128GB128GB410 tok/scomfortable
M4 Pro 48GB48GB205 tok/scomfortable
M4 Pro 24GB24GB205 tok/scomfortable
A100 PCIe 80 GB80GB2055 tok/scomfortable
H100 SXM5 80 GB80GB4134 tok/scomfortable
GeForce RTX 309024GB991 tok/scomfortable
Radeon RX 7900 XTX24GB820 tok/scomfortable
GeForce RTX 408016GB796 tok/scomfortable
Install CLI [email protected] Raw data · MIT · API data: live · HW/Cloud data: curated 2026-02-23 · v0.6.0