lfm2:24b-a2b
OLLAMA GGUFlfm2moe · 23.8B · Q4_K_M
May 24, 2026 · Intel Gen Intel® Core™ i7-1355U
qwen3.6:35b-a3b-q8_0
OLLAMA GGUFqwen35moe · 36.0B · Q8_0
Jul 7, 2026 · Cortex-X925
Global Score
78 vs 80
Hardware Fit
67 vs 79
Quality Score
82 vs 81
Hardware
lfm2:24b-a2b qwen3.6:35b-a3b-q8_0
MachineDell Inc. Inspiron 15 3530ASUSTeK COMPUTER INC. GX10
CPUIntel Gen Intel® Core™ i7-1355UCortex-X925
Cores1220
RAM31 GB122 GB
GPURaptor Lake-P [Iris Xe Graphics]Device 2e12
OScachyos rollingUbuntu 24.04.4 LTS
Archx64arm64
Power Modelow-powerperformance
Performance
lfm2:24b-a2b qwen3.6:35b-a3b-q8_0
Tokens/sec11.555.5
First chunk1727 ms528 ms
TTFT1.7 s30.0 s
Load time6.8 s32.6 s
Memory usage13.9 GB41.0 GB
Memory %45%34%
HW Fit Score Breakdown
lfm2:24b-a2b
Speed
22/50
TTFT
19/20
Memory
26/30
qwen3.6:35b-a3b-q8_0
Speed
50/50
TTFT
0/20
Memory
29/30
Quality
lfm2:24b-a2b
Reasoning
13/20
Coding
17/20
Instruction
16/20
Structured
15/15
Math
11/15
Multilingual
10/10
Reasoning: Adequate Coding: Strong Instruction Following: Strong Structured Output: Strong Math: Adequate Multilingual: Strong
qwen3.6:35b-a3b-q8_0
Reasoning
18/20
Coding
13/20
Instruction
17/20
Structured
10/15
Math
13/15
Multilingual
10/10
Reasoning: Strong Coding: Adequate Instruction Following: Strong Structured Output: Adequate Math: Strong 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