Skip to content

Head-to-head · LFM

LFM2 350M vs LFM2.5 8B-A1B

LFM2 350M needs ~1.4 GB at Q4_K_M; LFM2.5 8B-A1B needs ~6.7 GB. That ~5.3 GB gap decides which hardware runs each; they differ on 3 of the 14 devices below.

Spec LFM2 350M LFM2.5 8B-A1B
Parameters0.354B8.3B MoE
Memory at Q4_K_M~1.4 GB~6.7 GB
Context window128k128k
LMArenanot rankednot ranked
LicenseLFM Open License v1.0LFM Open License v1.0

Which devices run each

A representative spread across the memory range. Tap a verdict for the full breakdown.

Size at each quantization

Quant LFM2 350M LFM2.5 8B-A1B
Q2_K 0.1 GB* 3.5 GB*
Q3_K_M 0.2 GB* 4.1 GB*
Q4_K_M 0.45 GB 5.2 GB
Q5_K_M 0.3 GB* 5.9 GB*
Q6_K 0.3 GB* 6.8 GB*
Q8_0 0.35 GB 9 GB
FP16 0.711 GB 17 GB

* derived from bits-per-weight; unstarred sizes are measured GGUF files.

Bottom line

The larger LFM2.5 8B-A1B needs ~6.7 GB; LFM2 350M runs on lighter hardware at ~1.4 GB with more headroom and faster responses. Check each against your exact device: what runs LFM2 350M · what runs LFM2.5 8B-A1B.

More LFM comparisons

FAQ

What is the difference in memory between LFM2 350M and LFM2.5 8B-A1B?

At Q4_K_M, LFM2 350M needs about 1.4 GB and LFM2.5 8B-A1B needs about 6.7 GB, a difference of ~5.3 GB.

Should I run LFM2 350M or LFM2.5 8B-A1B?

Run LFM2.5 8B-A1B if your hardware has the ~6.7 GB it needs and you want maximum quality; run LFM2 350M (~1.4 GB) for lighter hardware, faster responses, and more memory headroom.

Full breakdowns: LFM2 350M · LFM2.5 8B-A1B · all models × devices.

Sources

Memory figures are estimates at Q4_K_M, validated 2026-08-03. See methodology.