laguna-xs-2.1
OLLAMA GGUFAug 1, 2026 · AMD RYZEN AI MAX+ 395
hf.co/lmstudio-community/gemma-4-E4B-it-GGUF:Q6_K
OLLAMA GGUFgemma4 · 7.52B · unknown
Aug 23, 2026 · Apple M1 Pro
Global Score
84 vs 88
Hardware Fit
100 vs 100
Quality Score
77 vs 83
Hardware
laguna-xs-2.1 hf.co/lmstudio-com…
MachineAZW GTR ProMacBook Pro
CPUAMD RYZEN AI MAX+ 395Apple M1 Pro
Cores3210
RAM125 GB16 GB
GPURadeon 8060SApple M1 Pro
OSUbuntu 24.04.4 LTSmacOS 26.6.2
Archx64arm64
Power Modelow-powerbalanced
Performance
laguna-xs-2.1 hf.co/lmstudio-com…
Tokens/sec65.934.6
First chunk281 ms210 ms
TTFT281 ms210 ms
Load time9.9 s4.0 s
Memory usage30.1 GB3.9 GB
Memory %24%25%
HW Fit Score Breakdown
laguna-xs-2.1
Speed
50/50
TTFT
20/20
Memory
30/30
hf.co/lmstudio-communit…
Speed
50/50
TTFT
20/20
Memory
30/30
Quality
laguna-xs-2.1
Reasoning
8/20
Coding
18/20
Instruction
16/20
Structured
14/15
Math
11/15
Multilingual
10/10
Reasoning: Weak Coding: Strong Instruction Following: Strong Structured Output: Strong Math: Adequate Multilingual: Strong
hf.co/lmstudio-communit…
Reasoning
14/20
Coding
18/20
Instruction
17/20
Structured
15/15
Math
9/15
Multilingual
10/10
Reasoning: Adequate Coding: Strong Instruction Following: Strong Structured Output: Strong Math: Adequate Multilingual: Strong
Run yours and compare
$
npm install -g metrillm@latest$
metrillmRequires Node 20+ and Ollama or LM Studio running
Or run without installing: npx metrillm@latest