On Wednesday, Google ne apne Bard chatbot mein ek nayi pidi ki artificial intelligence model Gemini ko joda hai. Isko Google ChatGPT aur dusre takkar ke rivals se bhi behtar tark karne ki kshamata hai. Ye search engine raja OpenAI ke ChatGPT-maker se generative AI mein aage badhne ka lakshya rakhta hai, jabki OpenAI apne boardroom coup ke natije ke sath samna kar raha hai, jisme chief executive Sam Altman ko pehle nikaala gaya aur phir dino mein wapas rakh liya gaya hai.
Gemini ek aisa AI model hai jaise GPT-4, jise seedhe taur par upyog nahin kiya jaa sakta hai. Balki, yeh Google aur ant mein dusre developers dwara istemaal ki jaane wali products banane ke liye ek base ke roop mein kaam karta hai. Google ka daava hai ki Gemini ko shuru se hi multimodal banaya gaya hai, iska matlab hai ki yeh alag alag prakar ke jankari jaise text, audio, image, code, aur video ko ek saath milakar aur combine karke chal sakta hai. Yeh tasveerein pehchaan sakta hai, real-time mein baat kar sakta hai, aur ajaib se bhi bhautik vigyan ka samadhan kar sakta hai. Sirf demo ko dekh lo.
Gemini 1.0 mein, teen models hain - Ultra, Pro aur Nano. Gemini Ultra Google ki sabse shaktishali LLM hai aur iska istemal "atyant complex tasks" ke liye enterprise applications mein kiya jayega. Gemini Pro teeno models mein sabse general-purpose hai aur ise pehle se hi Bard ke liye use kiya ja chuka hai jahan advanced reasoning, planning aur understanding ki zaroorat hoti hai.
Link: https://deepmind.google/technologies/gemini/#introduction
December 13 se onwards, developers aur enterprise customers Gemini API ke zariye Google AI Studio ya Google Cloud Vertex AI mein is model tak pahunch sakte hain. Wahi toh, Gemini Nano ko sabse efficient model mana gaya hai on-device tasks ke liye, aur isko Pixel 8 Pro mein integrate kiya gaya hai jisse information summarisation aur Smart Reply jaise tasks ko process kiya ho sake.
