qwen3-30b-a3b-thinking-2507-claude-4.5-sonnet-high-reasoning-distill-mlx

qwen3_moe · 30B · MXFP4

THINKING MODEL

Mac Studio (Apple M2 Max)

96 GB · macOS 26.1

Tested on April 22, 2026 · Submitted by Frank

Top 1% Compare

Global Score

95 /100

Excellent

Hardware Fit

94/100

Quality

96/100

Get this model

Get it in LM Studio

Search and download models directly from the app

🤗

Find on HuggingFace

MLX versions & conversions

Hardware

Machine: Mac Studio
CPU: Apple M2 Max
Cores: 12 total (8 perf + 4 eff)
Frequency: 2.4 GHz
RAM: 96 GB LPDDR5
GPU: Apple M2 Max
OS: macOS 26.1
Arch: arm64
Power Mode: balanced

Performance

Tokens/sec Tokens generated per second — higher is better: 80.4
Standard deviation Variation of speed across benchmark runs — lower means more consistent: ±0.7
First chunk latency Delay before the first streamed chunk arrives from runtime: 46 ms
Time to first token How fast the model starts responding — lower is better: 1.8 s
Load time Time to load the model into memory before inference: N/A
Memory usage RAM consumed during inference — percentage of total system memory: 21.2 GB (22%)
Total tokens Total number of tokens generated across all benchmark prompts: 1560
Thinking tokens (est.) Estimated internal reasoning tokens — models with "thinking" generate more tokens: ~449

Score breakdown

Speed

50/50

Time to first token

14/20

Memory

30/30

Quality

Reasoning

19/20

Coding

18/20

Instruction following

19/20

Structured output

15/15

Math

15/15

Multilingual

10/10

Category levels

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

Metadata

Spec version: 0.2.1
Runtime: LM Studio 0.4.12+1
Model format: MLX
Hardware profile: HIGH-END
Result hash: a4c08eb958cca2648f50c00ea6d751804e9becfb0b88be9d8dbee6fee980b4df

Interpretation

Hardware fit: 94/100. Overall suitability: EXCELLENT (Global 95/100). Category profile: Reasoning: Strong, Coding: Strong, Instruction Following: Strong, Structured Output: Strong, Math: Strong, Multilingual: Strong.

Warnings

Model memory footprint is estimated via LM Studio CLI rather than measured from a fresh load.

Bench Environment

Power: AC CPU load: avg 15% (peak 20%)

Run yours now

$ npm install -g metrillm@latest

$ metrillm

Requires Node 20+ and Ollama or LM Studio running

Or run without installing: npx metrillm@latest