Back to leaderboard

qwen/qwen3-30b-a3b-2507

LM-STUDIO MLX

qwen3_moe · 30B · 4bit

Excellent

Mar 4, 2026 · Apple M4

deepseek-coder-1.3b-instruct

LM-STUDIO GGUF

llama · 1.3B · Q8_0

Not Rec.

Mar 6, 2026 · Intel Core™ i5-5300U

Global Score
94 vs 30
Hardware Fit
100 vs 66
Quality Score
91 vs 14

Hardware

qwen/qwen3-30b-a3b… deepseek-coder-1.3…
MachineMacBook AirLENOVO 20BUS00700
CPUApple M4Intel Core™ i5-5300U
Cores104
RAM32 GB16 GB
GPUApple M4Intel(R) HD Graphics 5500
OSmacOS 26.3Microsoft Windows 10 Professionnel 10.0.19045
Archarm64x64
Power Modebalancedbalanced

Performance

qwen/qwen3-30b-a3b… deepseek-coder-1.3…
Tokens/sec44.87.1
First chunk405 ms3448 ms
TTFT405 ms3.4 s
Load timeN/AN/A
Memory usage16.0 GB1.3 GB
Memory %50%9%

HW Fit Score Breakdown

qwen/qwen3-30b-a3b-2507

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

deepseek-coder-1.3b-ins…

Speed
24/50
TTFT
15/20
Memory
27/30

Quality

qwen/qwen3-30b-a3b-2507

Reasoning
17/20
Coding
19/20
Instruction
17/20
Structured
15/15
Math
13/15
Multilingual
10/10
Reasoning: Strong Coding: Strong Instruction Following: Strong Structured Output: Strong Math: Strong Multilingual: Strong

deepseek-coder-1.3b-ins…

Reasoning
0/20
Coding
3/20
Instruction
3/20
Structured
6/15
Math
0/15
Multilingual
2/10
Reasoning: Poor Coding: Poor Instruction Following: Poor Structured Output: Weak Math: Poor Multilingual: Poor
View qwen/qwen3-30b-a3b… details | View deepseek-coder-1.3… details

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