What is the minimum hardware to run ERNIE 4.5 21B-A3B?

Nvidia GeForce RTX 4060 Ti (16GB) is the lightest device that runs ERNIE 4.5 21B-A3B, using ~14.3 GB at Q4_K_M of its ~15 GB usable. Anything with less usable memory cannot load it.

How much memory does ERNIE 4.5 21B-A3B need?

At Q4_K_M it needs about 14.3 GB (weights ~12.42 GB + KV cache + overhead at 4k context). At Q8_0 budget ~23.5 GB.

Can I run ERNIE 4.5 21B-A3B on a Mac?

Yes. ERNIE 4.5 21B-A3B runs on Apple Silicon with enough unified memory; the ranked list below shows the lightest Mac that fits.

Can I run ERNIE 4.5 21B-A3B on a phone?

No. ERNIE 4.5 21B-A3B is too large for current phone memory at Q4_K_M; stick to a laptop, desktop GPU, or Mac.

Hardware guide · ERNIE

ER What hardware do you need to run ERNIE 4.5 21B-A3B?

ERNIE 4.5 21B-A3B needs about 14.3 GB to run at Q4_K_M, so the lightest hardware that runs it is Nvidia GeForce RTX 4060 Ti (16GB). 17 of 40 devices tested can run it.

Needs (Q4_K_M): ~14.3 GB
Devices that run it: 17
Too small: 23
Lightest that runs it: ~15 GB

Recommended Q4_K_M

Apple M4 (24GB) Yes

Runs at Q4_K_M using ~14.3 GB of ~16 GB usable.

What kind of hardware runs ERNIE 4.5 21B-A3B

Of the 17 devices that run it, here is the split by hardware class.

7 discrete GPUs 9 Macs 1 laptop

Hardware that runs ERNIE 4.5 21B-A3B

Ranked by usable memory, lightest first. Prices are approximate street prices for the device itself (a GPU is the card alone; a Mac is the whole machine). Tok/s is a bandwidth estimate; see methodology.

Compatible devices 17 total

See the full ERNIE 4.5 21B-A3B memory breakdown, or compare it against other models.

Sources

Memory at Q4_K_M (14.3 GB = 12.42 GB weights + KV cache + overhead), validated 2026-08-03. See methodology.

ER What hardware do you need to run ERNIE 4.5 21B-A3B?

What kind of hardware runs ERNIE 4.5 21B-A3B

Hardware that runs ERNIE 4.5 21B-A3B

Too small for ERNIE 4.5 21B-A3B

Sources