BEIJING, China – Kamfanin fasahar AI na China, DeepSeek, ya fara zama babban abin tuhuma a fagen fasahar AI na duniya bayan fitar da sabbin samfuran AI masu inganci da aka yi amfani da su a matsayin albarkatu na buɗe tushe. Waɗannan samfuran suna ƙalubalantar manyan kamfanoni kamar OpenAI, Google, da Meta.
An kafa DeepSeek a watan Mayu 2023 ta hannun Wenfeng, wanda kuma ya kafa kamfanin zuba jari na High-Flyer. Kamfanin ya sami kuɗi ne kawai daga High-Flyer, wanda ya ba shi damar yin bincike mai zurfi ba tare da matsin lamba daga masu zuba jari ba. Ƙungiyar DeepSeek ta ƙunshi ƙwararrun ƙwararrun daliban jami’o’i na China, waɗanda suka ba da gudummawa ga ci gaban samfuran AI masu inganci.
A cikin Nuwamba 2023, DeepSeek ya fara fitar da samfurin AI na farko don ayyukan coding. Bayan haka, a cikin Mayu 2024, sun fitar da DeepSeek LLM, wanda ya haifar da yakin farashi a kasuwar AI ta China. Waɗannan samfuran sun tilasta wa manyan kamfanoni kamar ByteDance, Tencent, Baidu, da Alibaba rage farashin samfuran su.
DeepSeek ya ci gaba da fitar da samfuran AI masu inganci kamar DeepSeek-V2 da DeepSeek-V3, waɗanda ke da ƙarfin sarrafa bayanai masu rikitarwa. Samfurin DeepSeek-V3 yana da ƙarfin sarrafa bayanai har zuwa 128K tokens kuma yana da farashi mai rahusa ga masu amfani.
Duk da nasarorin da DeepSeek ya samu, kamfanin yana fuskantar ƙalubale da yawa, gami da ƙarancin kayan aikin lantarki da ƙuntatawa daga Amurka. Duk da haka, ci gaban da DeepSeek ya samu ya nuna cewa China na iya zama babbar ƙasa a fagen fasahar AI.
