EX

K EXAONE 236B A23B

ExaoneGGUF by mradermacherUpdated 2026-01-15
Download on Hugging Face
Parameters236B
Context256K
Layers49
Architectureexaone-moe
Smallest quant86.5 GB
Quick answer

K EXAONE 236B A23B needs about 145.6 GB of memory at Q4_K_M (a 143.5 GB file) with an 8K context window, which is more than a 48 GB GPU holds. This repository has 10 quants, from 86.5 GB (Q2_K) to 194.6 GB (Q6_K). Longer context needs more: at 32K tokens the KV cache alone adds about 6.6 GB.

236B multilingual MoE LLM with 23B active parameters

K EXAONE 236B A23B is a large‑scale multilingual language model built on an exaone‑moe architecture. It contains 236B total parameters, with 23B active during inference, and supports a 262,144‑token context window.

Good at

  • Long‑context reasoning
  • Agentic tool use
  • Multilingual understanding

Limits

  • Large VRAM requirement (≈88.6 GB for Q2_K)
  • No vision support
  • License: other

Best for

  • Processing long documents
  • Multilingual translation
  • Agentic task execution

Who it’s for. Developers and researchers needing high‑capacity multilingual models with long‑context and agentic capabilities. Suitable for applications that can accommodate the large memory footprint.

Quants & memory

QuantFileEst. memoryvs VRAMResultDownload
Q2_K86.5 GB 88.6 GB
Too big Get file →
Q3_K_S102.4 GB 104.5 GB
Too big Get file →
Q3_K_M113.3 GB 115.5 GB
Too big Get file →
Q3_K_L122.8 GB 125.0 GB
Too big Get file →
IQ4_XS127.7 GB 129.8 GB
Too big Get file →
Q4_K_S134.8 GB 136.9 GB
Too big Get file →
Q4_K_M143.5 GB 145.6 GB
Too big Get file →
Q5_K_S163.3 GB 165.4 GB
Too big Get file →
Q5_K_M168.3 GB 170.4 GB
Too big Get file →
Q6_K194.6 GB 196.8 GB
Too big Get file →

For 8 GB VRAM, the smallest quant is Q2_K at 86.5 GB, which still exceeds 8 GB. For a 16 GB VRAM budget, Q3_K_M (113.34 GB) is the next smallest. For a 32 GB VRAM budget, Q4_K_M (143.47 GB) provides a balance between size and performance. Memory estimates are approximate.

What GPU do you need for K EXAONE 236B A23B?

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: no quant fits at an 8K context.
  • 16 GB VRAM: no quant fits at an 8K context.
  • 24 GB VRAM: no quant fits at an 8K context.
  • 32 GB VRAM: no quant fits at an 8K context.
  • 48 GB VRAM: no quant fits at an 8K context.

Which K EXAONE 236B A23B quant should you choose?

Q4_K_M is the usual starting point for K EXAONE 236B A23B: it balances file size and quality well. Here it is a 143.5 GB file.

If memory is tight, the smallest file is Q2_K at 86.5 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

Architectureexaone-moe
Parameters236B (approximate, mixture of experts)
Active parameters23B
Layers49
Trained context262,144 tokens
KV cache (fp16)196 KB per token

Source

Quantized bymradermacher
FamilyExaone
Licenseother
Downloads (HF)1,565
Last updated2026-01-15

Frequently asked questions about K EXAONE 236B A23B

How much VRAM does K EXAONE 236B A23B need?

About 145.6 GB at Q4_K_M with an 8K context window. That is the 143.5 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 K EXAONE 236B A23B on an 8 GB GPU?

Not at an 8K context. The smallest file needs about 88.6 GB, so you would need a larger GPU or CPU offloading.

What is the file size of K EXAONE 236B A23B GGUF?

It depends on the quant: Q2_K is 86.5 GB, Q3_K_S is 102.4 GB, Q3_K_M is 113.3 GB, Q3_K_L is 122.8 GB, IQ4_XS is 127.7 GB, Q4_K_S is 134.8 GB and 4 more.

What context length does K EXAONE 236B A23B support?

The GGUF header lists a trained context length of 262,144 tokens (256K). The KV cache grows with the context you use: about 1.6 GB at 8K and 6.6 GB at 32K.

Where can I download K EXAONE 236B A23B GGUF?

The files are on Hugging Face in the mradermacher/K-EXAONE-236B-A23B-GGUF repository. The download buttons on this page open the file on Hugging Face. ggufmodels.com does not host any model files.

What license does K EXAONE 236B A23B use?

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

Can I run this model on 8 GB VRAM?

No. The smallest quant requires 86.6 GB of VRAM at 8K context.

Does this model support vision tasks?

No, it is a text‑only model.

What license does this model use?

The model is released under an other license.

Other GGUFs of K-EXAONE-236B-A23B

Similar models

Browse all →

More from mradermacher

All by mradermacher →

More Exaone models

All Exaone →

Other large models

All large models →