Leaderboard / LG AI Research (EXAONE)

EXAONE 4.0 32B

Name: EXAONE 4.0 32B
Author: LG AI Research (EXAONE)

LG AI Research (EXAONE) · EXAONE 4.0 · released 2025-07-15 · data checked 2026-11-01

EXAONE 4.0 32B is a 32B-parameter dense model from LG AI Research (EXAONE) (Rest of Asia), released 2025-07-15 under the EXAONE License (non-commercial) license. It scores 62.0 on the Small LM Index, #17 of 34 medium models. It needs about 21.9 GB of memory at 4-bit with an 8K-token context.

Small LM Index

62.0

#17 of 34 scored medium models

Likely 58.8–65.0 · from 4 of 6 core benchmarks

Parameters

32B

Dense

Price per 1M tokens

—

No hosted API on OpenRouter; self-host

Context window

131K

tokens

Benchmarks

Each bar shows this model's score. The tick marks the best medium model on that benchmark. Sources: 9 reported by LG AI Research (EXAONE). How we label sources.

MMLU-Pro Lab-reported Knowledge81.8

Class best: Qwen3.6 27B 86.2

GPQA Diamond Lab-reported Knowledge75.4

Class best: Nanbeige4.3 33B 89.5

AIME Lab-reported Math · AIME 202585.3

Class best: Ternary Bonsai 2 27B 95.8

LiveCodeBench Lab-reported Coding · LiveCodeBench v666.7

Class best: Qwen3.8 27B 90.3

MMLU Lab-reported Knowledge · MMLU-Redux92.3

Multilingual MMLU Lab-reported Multilingual · MMMLU Spanish only85.6

IFEval Lab-reported Instruction following83.7

Class best: LFM2.5 24B-A2B 96.8

BFCL Lab-reported Agentic · BFCL v363.9

Class best: Ternary Bonsai 2 27B 74.9

HMMT Lab-reported Math · HMMT Feb 202572.9

Class best: Granite 4.2 30B 89.2

Not reported: estimated

What the index model expects from this model's other scores. Not a measurement.

Humanity's Last Exam≈ 12

SWE-bench Verified≈ 49

Specifications

Lab: LG AI Research (EXAONE) (KR · Rest of Asia)
Architecture: Dense transformer
Total parameters: 32B
Active parameters: All (dense)
Size class: Medium
Context window: 131K tokens
Input modalities: text
Reasoning: Hybrid (thinking on/off)
License: EXAONE License (non-commercial)
Openness: Open weights
Memory to run: ~21.9 GB at 4-bit, 8K context (more sizes)
Hugging Face: 14.8K downloads (30 days) · 283 likes

Weights on Hugging Face ↗

Compare with

EXAONE 4.0 32B vs Nemotron 3.5 Lightning (31.6B, index 63.6)
EXAONE 4.0 32B vs Nemotron 3 Nano 30B-A3B (31.6B, index 62.6)
EXAONE 4.0 32B vs Sarvam 30B (32.2B, index 61.6)
EXAONE 4.0 32B vs OLMo 3.1 32B Think (32.2B, index 57.0)

Sources

https://huggingface.co/LGAI-EXAONE/EXAONE-4.0-32B

Something wrong here? Report it.

Run it locally

Uses LGAI-EXAONE/EXAONE-4.0-32B-GGUF (Q4_K_M). Swap the tag for another size: BF16, IQ4_XS, Q4_K_M, Q5_K_M, Q6_K, Q8_0.

Ollama

ollama run hf.co/LGAI-EXAONE/EXAONE-4.0-32B-GGUF:Q4_K_M

llama.cpp

llama-server -hf LGAI-EXAONE/EXAONE-4.0-32B-GGUF:Q4_K_M

Then open http://localhost:8080 for a chat window.

MLX (Mac)

pip install -U mlx-lm
mlx_lm.chat --model mlx-community/EXAONE-4.0-32B-4bit

vLLM (GPU server)

vllm serve LGAI-EXAONE/EXAONE-4.0-32B

Full-precision weights, for a server GPU.

Open in LM Studio

Memory needed: weights, context cache and runtime buffers.

Quantization	4K context	32K context	131K context
16-bit F16	66 GB	68 GB	75 GB
8-bit Q8_0	36 GB	38 GB	45 GB
4-bit Q4_K_M	22 GB	23 GB	30 GB

Check it against your GPU, Mac or phone →