True
Kakao has publicly released two AI vision encoder models developed in-house as open source, publishing them through its Kakao Tech channels, including a project image hosted on akaocdn.net. The move adds to the growing pool of vision models available to developers under open-source terms.
The two vision encoders were developed by Kakao itself and released as open source. Details of the models were published via Kakao Tech's official media channels. No specific benchmark figures, license terms, or model specifications were disclosed in the available materials.
Kakao, one of Korea's leading internet platforms, has been active in AI development, and open-sourcing its own vision encoders gives outside developers and researchers direct access to the company's in-house vision technology. The release follows a broader trend among Korean tech firms of contributing AI models to the open-source community rather than keeping them fully proprietary.
Kakao has not announced, in the materials reviewed, any specific performance targets or intended applications for the encoders.
Kakao's open-source release of its two self-developed AI vision encoders is true.