LFM2 24B-A2B Instruct

LFM2 24B-A2B Instruct loads in 14 GB at Q4_K_M and is comfortable from 32 GB of memory, Macs included from the MacBook Air M5 24GB. Below: the full memory math, the cheapest card that runs it well, and the fastest machine we track.

PARAMETERS
24B (2B active)
FORMAT
Q4_K_M
MIN MEMORY
24 GB
BEST FOR
Local AI agents, privacy-first tool calling, MCP workflows

Memory math

Weights (Q4_K_M)
14 GB
+ KV cache (16k)
~4.0 GB
Total at 16k
~18.0 GB
Comfortable from
32 GB

KV = fp16 estimate (q8_0 cache roughly halves it). "Comfortable" = weights + KV within the engine's tiered budget (~70-85% of memory).

Hardware snapshot

Cheapest GPU
~84 tok/s est. — ~$550 used
Fastest
~164 tok/s est.
Runs on a Mac?
~29 tok/s est.

Go deeper

More LFM2 models

Frequently asked questions

How much memory does LFM2 24B-A2B Instruct need?

14 GB for the Q4_K_M weights, plus ~4.0 GB of KV-cache at 16k context — about 18.0 GB total. Comfortable from 32 GB of VRAM or unified memory.

Does LFM2 24B-A2B Instruct run on a Mac?

Yes — from the MacBook Air M5 24GB (~29 tok/s est.). Unified memory means the RAM budget is the only limit.

What is the cheapest GPU for LFM2 24B-A2B Instruct?

The AMD Radeon RX 7900 XT is the cheapest tracked card that runs LFM2 24B-A2B Instruct comfortably — ~84 tok/s est. at ~$550 used.

Cite this page

ModelFit: LFM2 24B-A2B Instruct — specs, memory math and hardware verdicts.
https://modelfit.io/models/lfm2-24b-a2b/ (dataset updated 2026-09-03, CC BY 4.0).