Skip to content

Laguna XS 2.1 on M3 Ultra 96 GB: VRAM, speed and fit

Laguna XS 2.1 at 8-bit runs on the M3 Ultra 96 GB with 256K of context at about 86 tokens a second.

Text generation · Chat · Ollama / LM Studio

Runs comfortably · on M3 Ultra 96 GB · 8-bit (Q8_0) · Ollama / LM Studio · Chat

GPU memory
67.7 GB
of 96.0 GB
Context
256K
auto
Speed
86 tok/s
31 when full
First reply
1.7 s
2K prompt

Where the memory goes

Model
33.1 GB
Context
10.1 GB
Overhead
516 MB
OS
24.0 GB
Free
28.3 GB
Memory
67.7 GB of 96.0 GB
Context
256K tokens · auto
Speed
86 tok/s · 31 when full
First reply
1.7 s for a 2K prompt
Running cost
$3.55/mo at home · $66/mo on a cloud 2× RTX A5000

Validation: Validated

What Laguna XS 2.1 on M3 Ultra 96 GB was checked against
CheckTargetResultStatus
Fits in GPU memory72.0 GB usable43.7 GB neededPass
Headroom5% free28.3 GB free (39%)Pass
Context for chat16K or more256KPass
Writing speed15 tok/s or more86 tok/sPass
First replyunder 3 s, 2K prompt1.7 sPass

Targets: a reply you can read along with, a first reply the use case tolerates, and 5% of memory left free. Calculated from published specs, not benchmarked.

The setup

Model
Laguna XS 2.1
Parameters
33.4B, 3B active
Precision
8-bit (Q8_0)
Context
256K, auto
Chats at once
1
Use
Chat
Runs in
Ollama / LM Studio
Hardware
M3 Ultra 96 GB
GPU can use
72.0 GB of 96.0 GB

Running cost

At home
$3.55/mo
162 W at the wall
Renting instead
$66/mo
2× RTX A5000 at $0.54/h

4 h a day, power at $0.18 per kWh, cloud at RunPod on-demand prices as of Nov 22, 2026.

How to run it

llama.cpp
llama-server -hf mudler/Laguna-XS-2.1-APEX-GGUF
vLLM
vllm serve poolside/Laguna-XS-2.1
MLX (Mac)
mlx_lm.chat --model mlx-community/Laguna-XS-2.1-4bit

Laguna XS 2.1 on similar machines

All machines: Laguna XS 2.1 VRAM requirements. Everything that runs here: what can the M3 Ultra 96 GB run?

Other models that run on M3 Ultra 96 GB

Questions

Can Laguna XS 2.1 run on M3 Ultra 96 GB?

Laguna XS 2.1 at 8-bit runs on the M3 Ultra 96 GB with 256K of context at about 86 tokens a second.

Report VR-0OONQG8, calculated Nov 22, 2026 from published specs, not benchmarked.