EXAONE 3.5 32B Instruct AWQ
EXAONE 3.5 32B Instruct is LG AI Research's bilingual English/Korean instruction model, quantized to 4-bit AWQ for lower VRAM overhead. It supports a 32K context window. License is non-commercial only.
If you need a capable bilingual Korean/English model for internal research or prototyping, EXAONE 3.5 32B is a solid option — the AWQ quant makes a 32B model approachable on prosumer hardware. The hard blocker is the NC license: anything customer-facing or revenue-adjacent is off the table. The trust_remote_code requirement also warrants a closer look before deployment in any sensitive environment. Hedge: worth testing for non-commercial work, skip if you need a production path.
›Why this rating
Auto-generated rating (Opus 4.7 judge, claude-opus-4-7). Overall 9.45/10. License is correctly identified as EXAONE AI Model License 1.1 NC with commercialOk=false, matching the HF card's 'license: other' + EXAONE naming. Metadata (32B params, 32K context, LG AI Research vendor, bilingual EN/KO) all verifiable from the card. Editorial voice is honest and operator-grade — flags NC license, trust_remote_code, and even hedges on the likes/downloads ratio. Best use case is appropriately specific (non-commercial bilingual research/summarization). Weaknesses surface the right deployment traps. Family='other' is reasonable since EXAONE is its own architecture.
Overview
EXAONE 3.5 32B Instruct is LG AI Research's bilingual English/Korean instruction model, quantized to 4-bit AWQ for lower VRAM overhead. It supports a 32K context window. License is non-commercial only.
Strengths
- Bilingual English/Korean instruction tuning from a major Korean AI lab
- 32K context window handles long documents
- AWQ 4-bit quantization lowers VRAM requirements vs. full-precision 32B
- 31K+ HuggingFace downloads suggests reasonable real-world uptake
Weaknesses
- Non-commercial license (EXAONE AI Model License 1.1 NC) — no production or revenue-generating use
- Requires trust_remote_code due to custom architecture, which is a security consideration
- LG AI Research explicitly flags risk of inappropriate or biased outputs
- Low HF likes (17) relative to download count may indicate limited community validation
Quantization variants
Each quantization trades model quality for file size and VRAM. Q4_K_M is the most popular starting point.
| Quantization | File size | VRAM required |
|---|---|---|
| Q4_K_M | 17.6 GB | 23 GB |
Get the model
HuggingFace
Original weights
Source repository — direct quantization required.
Hardware that runs this
Cards with enough VRAM for at least one quantization of EXAONE 3.5 32B Instruct AWQ.
Models worth comparing
Same parameter band, plus what's one tier above and below — so you can decide what actually fits your hardware.
Frequently asked
What's the minimum VRAM to run EXAONE 3.5 32B Instruct AWQ?
Can I use EXAONE 3.5 32B Instruct AWQ commercially?
What's the context length of EXAONE 3.5 32B Instruct AWQ?
Source: huggingface.co/LGAI-EXAONE/EXAONE-3.5-32B-Instruct-AWQ
Reviewed by RunLocalAI Editorial. See our editorial policy for how we research and verify model claims.
Related — keep moving
Verify EXAONE 3.5 32B Instruct AWQ runs on your specific hardware before committing money.