Can OneJev 27B run on MacBook Pro M3 Pro 18GB?

Estimated weights fit · custom workflow

Q2_K · 4K context · formula estimate

What this means

OneJev 27B fits on the MacBook Pro M3 Pro · 18GB at Q2_K. We estimate it uses 13.9GB of the conservative 15GB working budget.

OneJev scores typed decisions in one forward pass. Generic decode estimates are not decision latency; image inputs need additional projector memory. Use the publisher's qev server and System One client.

View technical details ↓Setup guide →

Not measured

decision latency

13.9GB

needed at Q2_K

Q2_K

quant selected

1.1GB

working headroom

needs 13.9 GBworking budget 15 GB

Apple M3 Pro18 GB unified memory MetalActive cooling

Where the memory goes

Component	Detail	GB
Model weights	Q2_K GGUF (10.7 GB) + mmap overhead	11.2
KV cache	4K context window	1.9
Runtime	macOS inference app + compute buffers	0.8
Total needed	at Q2_K, 4K context	13.9
Working budget	18 GB unified memory − conservative macOS reserve	15
Headroom	remaining inside the working budget	1.1

Pick your quant

Quant	Download	Memory	Estimated speed	Verdict
Q2_K ★	10.7 GB	13.9 GB	~7–9.8 tok/s	! Runs, barely
Q3_K_M	13.3 GB	16.7 GB	—	✕ Won't fit
IQ4_XS	15.2 GB	18.7 GB	—	✕ Won't fit
Q4_K_S	15.6 GB	19.1 GB	—	✕ Won't fit
Q4_K_M	16.5 GB	20 GB	—	✕ Won't fit
Q5_K_M	19.2 GB	22.9 GB	—	✕ Won't fit
Q6_K	22.1 GB	25.9 GB	—	✕ Won't fit
Q8_0	28.6 GB	32.7 GB	—	✕ Won't fit

Use this model for its intended task

This is a task model, not a normal local chat download.

OneJev 27B is designed to return calibrated probabilities for typed choices from text or images. This Mac has enough working memory for the tracked Q2_K weights, but loading the file in a plain chat window does not supply the inputs, tools, or action loop that make the model useful.

Start with the publisher's model page.

It documents the intended workflow and supported developer runtimes. This is not yet a point-and-click setup for beginners, so do not install LM Studio expecting an ordinary chat tutorial to reproduce the model's task.

Open OmniJev/OneJev-27B ↗

Just want to chat locally?

Choose Llama 3.2 3B instead. Its page gives the complete graphical install, download, first prompt, and offline check.

This model is designed to return calibrated probabilities for typed choices from text or images. Its weights may fit in memory, but it needs the publisher's qev server and System One client; image inputs also need a vision projector. A normal LM Studio chat does not provide that workflow.

Other models on this Mac

Bonsai 27B (1-bit) · Q1_0 Ternary Bonsai 27B · Q2_0 Qwen3 14B · Q4_K_M Phi-4 14B · Q4_K_M Ministral 3 14B · Q4_K_M Gemma 3 12B · Q4_K_M

OneJev 27B on other MacBook Pro M3 Pro configurations

Check 36GB

FAQ

Can the MacBook Pro M3 Pro · 18GB run OneJev 27B?

The tracked Q2_K working memory is estimated at 13.9GB, excluding the vision projector. Weight fit is not verified workflow support; use the publisher's qev server and System One client.

Which OneJev 27B quant should I use on this Mac?

Q2_K. It is a 10.7GB download and leaves about 1.1GB inside our conservative working budget.

Which app should I use for OneJev 27B on this Mac?

OneJev 27B is a task-specific model, not a normal local chat download. The selected weights fit this Mac, but the model still needs its publisher's intended workflow. This page links that repository and recommends Llama 3.2 3B if you just want local chat.

Are these speeds measured on a MacBook Pro M3 Pro?

Decision latency has not been measured. OneJev scores typed choices in one forward pass; generic decode estimates do not measure this workflow.