Back to leaderboard

qwen2.5:72b

OLLAMA GGUF

qwen2 · 72.7B · Q4_K_M

Not Rec.

Jun 27, 2026 · Cortex-X925

qwen3.5-35b-a3b-t48-c8k:latest

OLLAMA GGUF

qwen35moe · 36.0B · Q4_K_M

Not Rec.

Aug 1, 2026 · Intel Xeon® E5-4620 0

Global Score
75 vs 65
Hardware Fit
50 vs 38
Quality Score
86 vs 76

Hardware

qwen2.5:72b qwen3.5-35b-a3b-t4…
MachineNVIDIA NVIDIA_DGX_SparkDell Inc. PowerEdge R820
CPUCortex-X925Intel Xeon® E5-4620 0
Cores2064
RAM122 GB256 GB
GPUDevice 2e12Microsoft Remote Display Adapter, Microsoft Basic Display Adapter
OSUbuntu 24.04.4 LTSMicrosoft Windows Server 2022 Datacenter 10.0.20348
Archarm64x64
Power Modeperformancebalanced

Performance

qwen2.5:72b qwen3.5-35b-a3b-t4…
Tokens/sec4.43.4
First chunk529 ms4210 ms
TTFT529 ms4.2 s
Load time49.7 s45.1 s
Memory usage54.0 GB22.0 GB
Memory %44%9%

HW Fit Score Breakdown

qwen2.5:72b

Speed
4/50
TTFT
20/20
Memory
26/30

qwen3.5-35b-a3b-t48-c8k…

Speed
3/50
TTFT
5/20
Memory
30/30

Quality

qwen2.5:72b

Reasoning
19/20
Coding
19/20
Instruction
12/20
Structured
14/15
Math
12/15
Multilingual
10/10
Reasoning: Strong Coding: Strong Instruction Following: Adequate Structured Output: Strong Math: Strong Multilingual: Strong

qwen3.5-35b-a3b-t48-c8k…

Reasoning
16/20
Coding
15/20
Instruction
9/20
Structured
15/15
Math
11/15
Multilingual
10/10
Reasoning: Strong Coding: Strong Instruction Following: Weak Structured Output: Strong Math: Adequate Multilingual: Strong

Run yours and compare

$ npm install -g metrillm@latest
$ metrillm

Requires Node 20+ and Ollama or LM Studio running

Or run without installing: npx metrillm@latest