Back to leaderboard

mlx-community/Llama-3.2-3B-Instruct-4bit

LM-STUDIO GGUF
Good

Mar 5, 2026 · Apple M4

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

OLLAMA GGUF

qwen35moe · 36.0B · Q4_K_M

Not Rec.

Sep 6, 2026 · Intel Xeon® E5-4650 v2

Global Score
74 vs 81
Hardware Fit
100 vs 63
Quality Score
63 vs 88

Hardware

mlx-community/Llam… qwen3.5-35b-a3b-t4…
MachineMacBook AirDell Inc. PowerEdge R820
CPUApple M4Intel Xeon® E5-4650 v2
Cores1080
RAM32 GB256 GB
GPUApple M4Microsoft Remote Display Adapter, NVIDIA Tesla P4, NVIDIA Tesla P4, Microsoft Basic Display Adapter
OSmacOS 26.3Microsoft Windows Server 2022 Datacenter 10.0.20348
Archarm64x64
Power Modebalancedbalanced

Performance

mlx-community/Llam… qwen3.5-35b-a3b-t4…
Tokens/sec50.617.2
First chunk257 ms845 ms
TTFT257 ms845 ms
Load timeN/A103.1 s
Memory usage0.7 GB21.9 GB
Memory %2%9%

HW Fit Score Breakdown

mlx-community/Llama-3.2…

Speed
50/50
TTFT
20/20
Memory
30/30

qwen3.5-35b-a3b-t48-c8k…

Speed
16/50
TTFT
17/20
Memory
30/30

Quality

mlx-community/Llama-3.2…

Reasoning
9/20
Coding
14/20
Instruction
14/20
Structured
15/15
Math
3/15
Multilingual
8/10
Reasoning: Weak Coding: Adequate Instruction Following: Adequate Structured Output: Strong Math: Poor Multilingual: Strong

qwen3.5-35b-a3b-t48-c8k…

Reasoning
17/20
Coding
18/20
Instruction
16/20
Structured
15/15
Math
12/15
Multilingual
10/10
Reasoning: Strong Coding: Strong Instruction Following: Strong Structured Output: Strong Math: Strong 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