qwen3:8b

qwen3 · 8.2B · Q4_K_M

Marginal

Mar 2, 2026 · Apple M4

deepseek-v2:16b

deepseek2 · 15.7B · Q4_0

Good

Mar 2, 2026 · Apple M4

Global Score

47 vs 69

Hardware Fit

67 vs 100

Quality Score

38 vs 55

Hardware

qwen3:8b deepseek-v2:16b

MachineMacBook AirMacBook Air

CPUApple M4Apple M4

Cores1010

RAM32 GB32 GB

GPUApple M4Apple M4

OSmacOS 26.3macOS 26.3

Archarm64arm64

Power Modebalancedbalanced

Performance

qwen3:8b deepseek-v2:16b

Tokens/sec18.756.6

TTFT12.1 s239 ms

Load time1.2 s1.2 s

Memory usage9.5 GB17.7 GB

Memory %30%55%

HW Fit Score Breakdown

qwen3:8b

Speed

31/40

TTFT

6/30

Memory

30/30

deepseek-v2:16b

Speed

40/40

TTFT

30/30

Memory

30/30

Quality

qwen3:8b

Reasoning

5/20

Coding

5/20

Instruction

4/20

Structured

12/15

Math

2/15

Multilingual

10/10

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

deepseek-v2:16b

Reasoning

7/20

Coding

9/20

Instruction

13/20

Structured

10/15

Math

8/15

Multilingual

8/10

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

View qwen3:8b details | View deepseek-v2:16b details

Run yours and compare

npx metrillm@latest bench

Requires Node 20+ and Ollama or LM Studio running