SK Telecom said on the 25th that it has upgraded its artificial intelligence (AI) service adot by introducing the latest AI models, including A.X 4.0 and GPT-5, and adding a text security feature.
SK Telecom added the latest global large language models (LLMs) to the adot search service. The newly introduced AI models are A.X 4.0, the latest version of A.X developed in-house by SK Telecom, and the GPT-5 series, OpenAI's latest high-performance AI model. A.X 4.0 is characterized by a high level of understanding of Korean and Korean culture.
According to evaluation results released in Jul. this year, A.X 4.0 scored 78.3 on the Korean Massive Multi-Task Language Understanding (KMMLU), a Korean proficiency metric, higher than GPT-4o (72.5). It also scored 83.5 on CLIcK, a Korean language and culture benchmark, higher than GPT-4o (80.2).
The company said the GPT-5 series can handle more complex tasks than previous models, has strengthened reasoning capabilities, and has a lower likelihood of hallucinations.
Users can freely use 19 models, including Claude, Perplexity, Gemini, and Liner, alongside A.X. 4.0 and the GPT-5 series.
In this update, a new text security feature, "AI Message," was added to the AI call service adot Phone. AI Message was developed to protect customers from threats such as spam and phishing. Using AI technology, it warns customers of risks by labeling messages suspected of spam or phishing among the texts they receive with "impersonation," "fraud," and "link caution."
Whether a message is suspected of spam or phishing is determined by an AI model that conducts a comprehensive analysis of various factors—such as the caller ID, the content of the text, and any included links (URLs)—based on the vast spam and phishing data accumulated in adot Phone.
AI Message is available in the adot Phone app on the Android operating system and in the call menu within the adot app on the iOS operating system. It is not supported in default messaging apps.