Models & assistants Model or AI service
SK Telecom adapts A.X 4.0 for Korean language tasks
SK Telecom released two A.X 4.0 language model sizes for Korean. The model card identifies Qwen2.5 as the base and describes further training on Korean data. [1][2]
- License
- The 72B checkpoint’s model card lists the Qwen License. [2]
What the model does
On July 3, 2025, SK Telecom announced a 72B standard model and a smaller 7B option. It published the model weights and documented evaluations for Korean tasks. [1][2]
Why it matters
Offering two sizes lets a team assess Korean text generation at different compute budgets. The results also give teams a starting point for checking whether the adaptation helps their task. [1][2]
Who can use it
Language researchers and application developers can download the standard model or evaluate the lighter version for situations with less hardware. [1][2]
Key dates
Ways to use it
Compare both variants on representative Korean documents and verify named entities and numbers. Make sure model size, version and license are recorded with the results. [2]
Model limitations
Model cards distinguish the 72B and 7B versions. The standard 72B checkpoint lists a Qwen license, so check terms for the exact model used. [1][2]
Sources and verification
Catalog evidence checked How we verify records
- Primary source SK Telecom’s July 3 announcement of A.X 4.0 standard and lightweight models
- Primary source skt/A.X-4.0
- Other source skt/A.X-4.0 model card
- Primary source Hugging Face registry metadata for skt/A.X-4.0
Related profiles
Models & assistants
GPT-6.2 Astra and Sol
OpenAI
Astra targets demanding projects. Sol targets everyday tasks. Both support text and image input in the API.
Models & assistants
MAI Code 2 and Code 2 Flash
Microsoft AI
Code 2 targets complex software work. Flash targets frequent edits, explanations and routine fixes.
Models & assistants
MAI Realtime 1
Microsoft AI
Users can try a MAI Labs demo. Developers can register for a limited Foundry preview.