[whatmodel.sh](../../README.md) / K-EXAONE (236B-A23B)
# K-EXAONE (236B-A23B)
LG AI Research · released 2025-12 · 236B params (23B active) · k-exaone
**score 64.4** ±0.9 · #151 of 732 · from 25 benchmark results across 3 sources
Score within the [LMArena (text)](https://whatmodel.sh/?set=lmarena-text&score=set) comparison set only.
## What it's good at
● marks K-EXAONE (236B-A23B) among every model on that task or test, worst to best.
task vs the others rank counts tests
agent tasks ▁▁▃▆█▄●▅▄▄▂▁ #131 of 313 24% 7
coding ▁▁▁▃▅▇█▇●▄▃▁ #132 of 469 13% 3
chat & writing ▁▁▁▃▆▄▆█●▇▇▃ #138 of 402* 4.1% 1
math ▁▁▃▅█▆▄●▄▂▁▁ #105 of 431 31% 5
web search ▁●▁▅▄█▃▄·▁·▁ #80 of 84* 3.5% 1
agent tasks · #131 of 313 · 24% of its score
Beats [o3](o3.md), [Gemini 2.5 Pro](gemini-2-5-pro.md), [Gemini 3.1 Flash-Lite](gemini-3-1-flash-lite.md) here, though they rank higher overall. Trails [SmolLM3 3B](smollm3-3b.md), [Jamba 1.5 Large](jamba-1-5-large.md), [Devstral 2 (2512)](devstral-2.md) here, though they rank lower overall.
test result vs the others in its run difficulty counts why
SWE-bench Verified (maker) 49.4% ▁▁▁▁▁▁▂●█▆▁▁ beats 41/138 hard · avg 19% 5.9% sharp · at its level · small run
τ²-bench Telecom (maker) 73.5% ▁·▅█▆▅●▃▃▂▃▁ beats 58/89 hard · avg 17% 4.4% sharp · at its level · noisy · small run
Terminal-Bench 2.1 (maker) 30.3% ▁·▁·▁▁●▃▆█▅█ beats 10/63 very hard · avg 11% 3.9% sharp · at its level · small run
K-BrowseComp-Verified (Korean web… 10.3% ▃····●··▃█·▃ beats 1/8 very hard · avg 3.6% 3.5% typical
τ²-bench Retail (maker) 78.6% ▁····▂▁▃▆█●▁ beats 40/44 medium · avg 45% 2.7% at its level · small run
τ²-bench Airline (maker) 60.4% ▁▁·▂▁▃▂▁█●▂▃ beats 32/42 hard · avg 31% 2.3% at its level · small run
τ³-bench Banking (maker) 14.2% ▂▂·▅█▂▃●▃▆▂▅ beats 13/24 very hard · avg 6.1% 1.2% blunt · small run
coding · #132 of 469 · 13% of its score
Beats [Qwen3.5 Plus](qwen3-5-plus.md), [Qwen3-Coder-Next (80B-A3B)](qwen3-coder-next.md), [Gemini 2.5 Pro](gemini-2-5-pro.md) here, though they rank higher overall. Trails [Llama 3.1 Nemotron Ultra 253B v1](llama-3-1-nemotron-ultra-253b.md), [Seed-OSS 36B Instruct](seed-oss-36b.md), [Solar Pro 4](solar-pro-4.md) here, though they rank lower overall.
test result vs the others in its run difficulty counts why
SWE-bench Verified (maker) 49.4% ▁▁▁▁▁▁▂●█▆▁▁ beats 41/138 hard · avg 19% 5.9% sharp · at its level · small run
LiveCodeBench Pro 2025Q2 (Medium)… 25.9% █··▄▂▂██▄·▄● beats 21/22 very hard · avg 3.3% 3.9% sharp · small run
LiveCodeBench v6 (window unspecif… 80.7% ▂▃▄▄▅██▆●▇▄▆ beats 46/67 medium · avg 46% 3.4% small run
chat & writing · #138 of 402 · 4.1% of its score
Beats [o3-mini](o3-mini.md) here, though they rank higher overall. Trails [LFM2.5 8B-A1B](lfm2-5-8b-a1b.md), [LFM2.5 2.6B](lfm2-5-2-6b.md), [Nemotron-Cascade-8B](nemotron-cascade-8b.md) here, though they rank lower overall.
test result vs the others in its run difficulty counts why
IFEval (aggregation unspecified) … 89.7% ▄·▄·▄·▄▄···● beats 6/6 easy · avg 80% 4.1% several runs
math · #105 of 431 · 31% of its score
Beats [Claude Sonnet 4.6](claude-sonnet-4-6.md), [MiniMax-M3](minimax-m3.md), [Nemotron 3 Ultra 550B-A55B](nemotron-3-ultra-550b-a55b.md) here, though they rank higher overall. Trails [Gemini 2.5 Flash](gemini-2-5-flash.md), [GLM-4.5-Air](glm-4-5-air.md), [Granite 4.2 30B](granite-4-2-30b.md) here, though they rank lower overall.
test result vs the others in its run difficulty counts why
AIME 2026 (maker) 92.2% ▁▁▁▂▂▂▆●▆▂·▁ beats 30/52 medium · avg 61% 8.1% sharp · small run
IMO-AnswerBench (maker) 76.3% ▁▁·▁▁▁▄●█▄▂▃ beats 17/44 medium · avg 38% 8.0% small run · several runs
HMMT February 2026 (maker) 80.7% ▃▃▂▅·██●█▅▃▅ beats 20/35 hard · avg 35% 7.4% sharp · small run
AIME 2025 (maker) 92.8% █·▄▄····▄··● beats 5/6 medium · avg 54% 5.4% sharp
HMMT November 2025 (maker) 86.8% ▁▁····▁▅●█▇▁ beats 8/23 easy · avg 66% 2.0% small run
web search · #80 of 84 · 3.5% of its score
Trails [GPT-4.1 nano](gpt-4-1-nano-2025-04-14.md), [GPT-4o mini (2024-07-18)](gpt-4o-mini-2024-07-18.md), [Mistral Small 3.1 24B (2503)](mistral-small-3-1.md) here, though they rank lower overall.
test result vs the others in its run difficulty counts why
K-BrowseComp-Verified (Korean web… 10.3% ▃····●··▃█·▃ beats 1/8 very hard · avg 3.6% 3.5% typical
other tests (4)
test result vs the others in its run difficulty counts why
GPQA Diamond (maker) 79.1% █···▄▄···▄·● beats 5/6 hard · avg 51% 15% sharp · several runs
MMLU-Pro (maker) 83.8% ▁▁·▁▁▁▃▂▄▅●▁ beats 159/188 medium · avg 55% 8.1% small run · several runs
Humanity's Last Exam (maker) 13.6% ▁▁▁▄▆●▇██▆▁▃ beats 33/111 very hard · avg 4.3% 6.8% small run · several runs
IFBench (aggregation unspecified)… 67.3% ▂▆▂▄▄▄▂▄●▆█▇ beats 52/86 medium · avg 41% 3.6% at its level · small run · several runs
source results share of the score
LGAI-EXAONE/K-EXAONE-2.0-750B-A37B mode… 11 50% ▇▇▇▇▇▇
LG AI Research vendor-reported benchmar… 13 46% ▇▇▇▇▇▇
K-BrowseComp: A Web Browsing Agent Benc… 1 4% ▇
prior: pull toward the middle 0% ▇
weighted average = 64.4
Licences: LGAI-EXAONE/K-EXAONE-2.0-750B-A37B model card ([Apache-2.0](https://huggingface.co/LGAI-EXAONE/K-EXAONE-2.0-750B-A37B/blob/fec0c8dd25fdd20ffd09869d5af5277966d5523d/LICENSE)) · K-BrowseComp: A Web Browsing Agent Benchmark Grounded in Korean Contexts ([CC BY 4.0](https://creativecommons.org/licenses/by/4.0/))
## How to run it
### llama.cpp (recommended)
1. Install llama.cpp. Homebrew and winget installs name the commands llama-cli and llama-server instead. [Other ways to install](https://github.com/ggml-org/llama.cpp/blob/master/docs/install.md)
```sh
# macOS / Linux
curl -LsSf https://llama.app/install.sh | sh
# Windows
irm https://llama.app/install.ps1 | iex
```
2. Run K-EXAONE (236B-A23B) (Q4_K_M, needs ~143.47 GB, from [huggingface.co/LGAI-EXAONE](https://huggingface.co/LGAI-EXAONE/K-EXAONE-236B-A23B-GGUF/tree/5bd0394e4f42c00df63e207b9c434387523a6b77))
```sh
llama cli -hf LGAI-EXAONE/K-EXAONE-236B-A23B-GGUF:Q4_K_M
```
3. Or start a local server and open the address it prints (http://localhost:9931)
```sh
llama serve -hf LGAI-EXAONE/K-EXAONE-236B-A23B-GGUF:Q4_K_M
```
### LM Studio
1. Install LM Studio. Open the app once before using the lms command. [Download it (macOS / Windows / Linux)](https://lmstudio.ai/download)
2. Run K-EXAONE (236B-A23B) (Q4_K_M, needs ~143.47 GB, from [huggingface.co/LGAI-EXAONE](https://huggingface.co/LGAI-EXAONE/K-EXAONE-236B-A23B-GGUF/tree/5bd0394e4f42c00df63e207b9c434387523a6b77))
```sh
lms get https://huggingface.co/LGAI-EXAONE/K-EXAONE-236B-A23B-GGUF@Q4_K_M -y
```
3. Then run lms ls to see its name, and lms chat followed by that name
Other versions: 8bit (~251.4 GB, Mac only) `lms get https://huggingface.co/mlx-community/K-EXAONE-236B-A23B-8bit -y`
### MLX (Mac)
1. Install MLX (Mac). Apple silicon Macs only, text only.
```sh
# with uv
uv tool install mlx-lm
# or
pip install mlx-lm
```
2. Run K-EXAONE (236B-A23B) (8bit, needs ~251.4 GB, Mac only, from [huggingface.co/mlx-community](https://huggingface.co/mlx-community/K-EXAONE-236B-A23B-8bit/tree/c09f6395a23fc0a9e92e10569a3e2f2ea3fb3c5e))
```sh
mlx_lm.chat --model mlx-community/K-EXAONE-236B-A23B-8bit
```
3. Or start an OpenAI-compatible server at http://localhost:8080/v1
```sh
mlx_lm.server --model mlx-community/K-EXAONE-236B-A23B-8bit
```
### Other ways to run it
1. Run it as published (BF16, needs ~474.2 GB, NVIDIA only, from [huggingface.co/LGAI-EXAONE](https://huggingface.co/LGAI-EXAONE/K-EXAONE-236B-A23B/tree/61e6d578eb102b578e5704e2916ac841df9eca0a))
```sh
vllm serve LGAI-EXAONE/K-EXAONE-236B-A23B --reasoning-parser deepseek_v3 --tensor-parallel-size 4 --enable-auto-tool-choice --tool-call-parser hermes
```
Model repository and documentation: [huggingface.co/LGAI-EXAONE](https://huggingface.co/LGAI-EXAONE/K-EXAONE-236B-A23B)
[interactive version](https://whatmodel.sh/model/k-exaone-236b-a23b) · [JSON](https://whatmodel.sh/api?format=json&all=1)