llama-3.2-3b-instruct

llama · 3B · 4bit

Good

Mar 3, 2026 · Apple M4

qwen3.5-2b-mlx

qwen3_5 · 2B · 8bit

Marginal

Mar 3, 2026 · Apple M4

Global Score

72 vs 49

Hardware Fit

100 vs 83

Quality Score

60 vs 35

Hardware

llama-3.2-3b-instr… qwen3.5-2b-mlx

MachineMacBook AirMacBook Air

CPUApple M4Apple M4

Cores1010

RAM32 GB32 GB

GPUApple M4Apple M4

OSmacOS 26.3macOS 26.3

Archarm64arm64

Power Modebalancedbalanced

Performance

llama-3.2-3b-instr… qwen3.5-2b-mlx

Tokens/sec48.218.6

TTFT265 ms2.9 s

Load time0.0 s0.0 s

Memory usage0.0 GB0.0 GB

Memory %0%0%

HW Fit Score Breakdown

llama-3.2-3b-instruct

Speed

40/40

TTFT

30/30

Memory

30/30

qwen3.5-2b-mlx

Speed

31/40

TTFT

22/30

Memory

30/30

Quality

llama-3.2-3b-instruct

Reasoning

9/20

Coding

13/20

Instruction

14/20

Structured

13/15

Math

3/15

Multilingual

8/10

Reasoning: Weak Coding: Adequate Instruction Following: Adequate Structured Output: Strong Math: Poor Multilingual: Strong

qwen3.5-2b-mlx

Reasoning

0/20

Coding

9/20

Instruction

10/20

Structured

4/15

Math

3/15

Multilingual

9/10

Reasoning: Poor Coding: Weak Instruction Following: Adequate Structured Output: Weak Math: Poor Multilingual: Strong

View llama-3.2-3b-instr… details | View qwen3.5-2b-mlx details

Run yours and compare

npx metrillm@latest bench

Requires Node 20+ and Ollama or LM Studio running