deepseek-coder:6.7b

llama · 7B · Q4_0

MacBook Air (Apple M4)

32 GB · macOS 26.3

Tested on March 2, 2026

Top 68% Compare

Global Score

63 /100

Good

Hardware Fit

98/100

Quality

39/100

Get this model

🦙

Ollama

ollama pull deepseek-coder:6.7b

View on Ollama Library

ollama.com/library/deepseek-coder

🤗

Find on HuggingFace

GGUF versions & quantizations

Hardware

Machine: MacBook Air
CPU: Apple M4
Cores: 10 total (4 perf + 6 eff)
Frequency: 2.4 GHz
RAM: 32 GB LPDDR5
GPU: Apple M4
OS: macOS 26.3
Arch: arm64
Power Mode: balanced

Performance

Tokens/sec Tokens generated per second — higher is better: 24.1
Standard deviation Variation of speed across benchmark runs — lower means more consistent: ±0.4
Time to first token How fast the model starts responding — lower is better: 251 ms
Load time Time to load the model into memory before inference: 0.9 s
Memory usage RAM consumed during inference — percentage of total system memory: 12.6 GB (39%)
Total tokens Total number of tokens generated across all benchmark prompts: 1209

Score breakdown

Speed

38/40

Time to first token

30/30

Memory

30/30

Quality

Reasoning

4/20

Coding

14/20

Instruction following

9/20

Structured output

6/15

Math

1/15

Multilingual

5/10

Category levels

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

Metadata

Spec version: 0.2.0
Runtime: Ollama 0.17.4
Model format: GGUF
Hardware profile: BALANCED
Result hash: 0ea0d60f7c3e97b6cc713f1d27c7c5fa7d41a9335639dda31e08341149345e4e

Interpretation

Hardware fit: 98/100. Overall suitability: GOOD (Global 63/100). Category profile: Reasoning: Poor, Coding: Adequate, Instruction Following: Weak, Structured Output: Weak, Math: Poor, Multilingual: Weak.

Run yours now

npx metrillm@latest bench

Requires Node 20+ and Ollama running