Laguna S 2.1

Laguna S 2.1 loads in 96 GB at Q4_K_M and is comfortable from 128 GB of memory, Macs included from the MacBook Pro M5 Max 128GB. Below: the full memory math, the cheapest card that runs it well, and the fastest machine we track.

PARAMETERS
118B (8B active)
FORMAT
Q4_K_M
MIN MEMORY
128 GB
BEST FOR
Agentic coding, Long-horizon tasks

Memory math

Weights (Q4_K_M)
96 GB
+ KV cache (16k)
~6.0 GB
Total at 16k
~102.0 GB
Comfortable from
128 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
~10 tok/s est. — ~$3,847 used (as of 2026-07-31)
Fastest
~32 tok/s est.
Runs on a Mac?
~22 tok/s est.

More Laguna models

Frequently asked questions

How much memory does Laguna S 2.1 need?

96 GB for the Q4_K_M weights, plus ~6.0 GB of KV-cache at 16k context — about 102.0 GB total. Comfortable from 128 GB of VRAM or unified memory.

Does Laguna S 2.1 run on a Mac?

Yes — from the MacBook Pro M5 Max 128GB (~22 tok/s est.). Unified memory means the RAM budget is the only limit.

What is the cheapest GPU for Laguna S 2.1?

The AMD Ryzen AI Max+ 395 (Strix Halo) is the cheapest tracked card that runs Laguna S 2.1 comfortably — ~10 tok/s est. at ~$3,847 used (as of 2026-07-31).

Cite this page

ModelFit: Laguna S 2.1 — specs, memory math and hardware verdicts.
https://modelfit.io/models/laguna-s-2.1/ (dataset updated 2026-09-03, CC BY 4.0).