SK Telecom Releases A.X-K2 Multilingual LLM
The Korean telecom carrier's latest open language model targets English, Korean, Chinese, Japanese, and Spanish under a permissive license.

SK Telecom has published A.X-K2, a new entry in its A.X family of language models, on Hugging Face. The model is a text-focused system built for multilingual use, covering English, Korean, Chinese, Japanese, and Spanish.
The release carries an Apache-2.0 license, which allows commercial use, modification, and redistribution with minimal restrictions. That choice matters for developers and enterprises that want to build on the model without navigating the bespoke community terms that accompany many other open-weight releases.
Why it matters
SKT is one of a growing number of telecom operators investing directly in foundation models, and its A.X line reflects a regional bet: strong Korean-language performance paired with broader multilingual coverage. A permissively licensed model from a major carrier gives the Korean and wider Asian developer ecosystem another practical option beyond the usual US and Chinese labs.
Key details from the record:
- Company: SK Telecom
- Modality: text-based large language model
- Languages: English, Korean, Chinese, Japanese, Spanish
- License: Apache-2.0
SKT has not published parameter counts, context length, or benchmark figures in the available metadata, so the model's scale and positioning against peers remain to be seen. Interested developers can find the weights and any accompanying documentation on the official Hugging Face page.
Sources
- Visit
skt/A.X-K2
Hugging Face
More in Text / LLM
Meituan Ships a Lighter, Sparser LongCat-Flash
The food-delivery giant's newest open model trims its mixture-of-experts design for more efficient inference under an MIT license.
DeepSeek Refreshes V4-Flash With New 0731 Checkpoint
The MIT-licensed mixture-of-experts model returns in an updated build shipping with FP8 weights for cheaper inference.
DeepSeek Ships V4-Flash, a 304B MoE Tuned for Agents
The latest checkpoint in DeepSeek's V4 line leans into agentic workflows while keeping the permissive MIT license.
0 comments
No comments yet. Be the first to weigh in.