ER

Ernie-4.5-21B-A3B-Thinking

ErnieGGUF by lovedheartUpdated 2025-09-13
Download on Hugging Face
Parameters21.8B
Context128K
Layers28
Architectureernie4_5-moe
Smallest quant7.7 GB
Quick answer

Ernie-4.5-21B-A3B-Thinking needs about 10.1 GB of memory at IQ1_M (a 9.1 GB file) with an 8K context window, so it fits on a 12 GB GPU. This repository has 5 quants, from 7.7 GB (IQ1_S) to 14.2 GB (IQ4_XS). Longer context needs more: at 32K tokens the KV cache alone adds about 1.9 GB.

Ernie-4.5-21B-A3B-Thinking is a 21.8B-parameter Ernie model in GGUF format, quantized by lovedheart. Its trained context length is 128K tokens. The repository has 5 quant files, from 7.7 GB (IQ1_S) to 14.2 GB (IQ4_XS).

Quants & memory

QuantFileEst. memoryvs VRAMResultDownload
IQ1_S7.7 GB 8.7 GB
Too big Get file →
IQ2_XXS9.0 GB 10.0 GB
Too big Get file →
IQ1_M9.1 GB 10.1 GB
Too big Get file →
IQ3_XXS10.7 GB 11.7 GB
Too big Get file →
IQ4_XS14.2 GB 15.2 GB
Too big Get file →

What GPU do you need for Ernie-4.5-21B-A3B-Thinking?

These figures add the file size, an fp16 KV cache at an 8K context window (worked out from this model’s own header) and about 0.5 GB for the runtime. They are estimates, so leave a little headroom, and expect to use less context or a smaller quant if you are right at the limit.

  • 4 GB VRAM: no quant fits at an 8K context.
  • 6 GB VRAM: no quant fits at an 8K context.
  • 8 GB VRAM: no quant fits at an 8K context.
  • 12 GB VRAM: the best quant that fits is IQ3_XXS (10.7 GB file, about 11.7 GB in use).
  • 16 GB VRAM: the best quant that fits is IQ4_XS (14.2 GB file, about 15.2 GB in use).

Which Ernie-4.5-21B-A3B-Thinking quant should you choose?

IQ1_M (9.1 GB) is the middle option in this repository. Use it only when memory is very tight. Expect a clear drop in quality, which is usually worse on small models and easier to live with on large ones.

If memory is tight, the smallest file is IQ1_S at 7.7 GB, with the quality trade-offs that come with fewer bits. Use it only when memory is very tight. Expect a clear drop in quality, which is usually worse on small models and easier to live with on large ones.

New to quant names? Read GGUF quantization explained.

Specifications

Architecture from the GGUF header

Architectureernie4_5-moe
Parameters21.8B (approximate, mixture of experts)
Active parameters3B
Layers28
Trained context131,072 tokens
KV cache (fp16)56 KB per token

Source

Quantized bylovedheart
FamilyErnie
Licenseapache-2.0
Downloads (HF)15
Last updated2025-09-13

Frequently asked questions about Ernie-4.5-21B-A3B-Thinking

How much VRAM does Ernie-4.5-21B-A3B-Thinking need?

About 10.1 GB at IQ1_M with an 8K context window. That is the 9.1 GB file, plus the KV cache and roughly 0.5 GB of runtime overhead. Smaller quants need less and longer context needs more.

Can I run Ernie-4.5-21B-A3B-Thinking on an 8 GB GPU?

Not at an 8K context. The smallest file needs about 8.7 GB, so you would need a larger GPU or CPU offloading.

What is the file size of Ernie-4.5-21B-A3B-Thinking GGUF?

It depends on the quant: IQ1_S is 7.7 GB, IQ2_XXS is 9.0 GB, IQ1_M is 9.1 GB, IQ3_XXS is 10.7 GB, IQ4_XS is 14.2 GB.

What context length does Ernie-4.5-21B-A3B-Thinking support?

The GGUF header lists a trained context length of 131,072 tokens (128K). The KV cache grows with the context you use: about 0.5 GB at 8K and 1.9 GB at 32K.

Where can I download Ernie-4.5-21B-A3B-Thinking GGUF?

The files are on Hugging Face in the lovedheart/ERNIE-4.5-21B-A3B-Thinking-GGUF-IQ1_S repository. The download buttons on this page open the file on Hugging Face. ggufmodels.com does not host any model files.

What license does Ernie-4.5-21B-A3B-Thinking use?

Hugging Face lists the license as apache-2.0. Licenses can differ between the quantized files and the original model, so check the repository and the base model before commercial use.

Other GGUFs of ERNIE-4.5-21B-A3B-Thinking

Similar models

Browse all →

More from lovedheart

All by lovedheart →

More Ernie models

All Ernie →

Other medium models

All medium models →