India \u092e\u0947\u0902 AI \u0915\u0940 \u092c\u093e\u0924 \u0905\u0915\u094d\u0938\u0930 \u092c\u0921\u093c\u0947 foreign models \u092a\u0930 \u0939\u094b\u0924\u0940 \u0939\u0948, \u091c\u094b mostly English \u092a\u0930 train \u0939\u0941\u090f \u0939\u0948\u0902\u0964 \u0932\u0947\u0915\u093f\u0928 28 October 2026 \u0915\u094b \u090f\u0915 \u0916\u092c\u0930 \u0906\u0908 \u091c\u094b India \u0915\u0947 \u0905\u092a\u0928\u0947 AI ecosystem \u0915\u0947 \u0932\u093f\u090f important \u0939\u0948\u0964 Gurugram \u0915\u0940 company CoRover.ai \u0928\u0947 \u0905\u092a\u0928\u0947 3 BharatGPT models \u0915\u094b AIKosh platform \u092a\u0930 list \u0915\u0930 \u0926\u093f\u092f\u093e \u0939\u0948\u0964
AIKosh \u0915\u094d\u092f\u093e \u0939\u0948, \u092f\u0939 \u0938\u092e\u091d\u0928\u093e \u092a\u0939\u0932\u0947 \u091c\u0930\u0942\u0930\u0940 \u0939\u0948\u0964 AIKosh \u0938\u0930\u0915\u093e\u0930 \u0915\u093e National AI Datasets and Models Platform \u0939\u0948, \u091c\u094b IndiaAI Mission \u0915\u0947 under \u091a\u0932\u093e\u092f\u093e \u091c\u093e\u0924\u093e \u0939\u0948\u0964 \u0907\u0938 platform \u0915\u093e \u0915\u093e\u092e \u0939\u0948 India \u092e\u0947\u0902 \u092c\u0928\u0947 datasets \u0914\u0930 models \u0915\u094b \u090f\u0915 \u091c\u0917\u0939 \u092a\u0930 \u0932\u093e\u0928\u093e, \u0924\u093e\u0915\u093f developers, researchers \u0914\u0930 companies \u0909\u0928 \u0924\u0915 \u0906\u0938\u093e\u0928\u0940 \u0938\u0947 \u092a\u0939\u0941\u0902\u091a \u0938\u0915\u0947\u0902\u0964 CoRover \u0915\u0947 \u092e\u0941\u0924\u093e\u092c\u093f\u0915 \u0907\u0938 listing \u0938\u0947 \u092a\u0939\u0932\u0947 IndiaAI \u0914\u0930 Ministry of Electronics and Information Technology, \u092f\u093e\u0928\u0940 MeitY, \u0915\u0940 review, testing \u0914\u0930 approval process \u092a\u0942\u0930\u0940 \u0915\u0940 \u0917\u0908 \u0939\u0948\u0964<\/p>\n\n\n\n
BharatGPT CoRover.ai \u0915\u093e AI model family \u0939\u0948, \u091c\u094b India \u0915\u0940 languages \u0914\u0930 \u091c\u0930\u0942\u0930\u0924\u094b\u0902 \u0915\u0947 \u0932\u093f\u090f \u092c\u0928\u093e\u092f\u093e \u0917\u092f\u093e \u0939\u0948\u0964 \u0907\u0938 family \u0915\u093e focus \u0939\u0948 Indian languages \u092e\u0947\u0902 text \u0914\u0930 voice \u0926\u094b\u0928\u094b\u0902 \u0915\u094b \u0938\u092e\u091d\u0928\u093e, \u0924\u093e\u0915\u093f \u0932\u094b\u0917 \u0905\u092a\u0928\u0940 \u092d\u093e\u0937\u093e \u092e\u0947\u0902 AI \u0938\u0947 \u092c\u093e\u0924 \u0915\u0930 \u0938\u0915\u0947\u0902\u0964 AIKosh \u092a\u0930 list \u0939\u0941\u090f 3 models \u0905\u0932\u0917-\u0905\u0932\u0917 \u0915\u093e\u092e \u0915\u0947 \u0932\u093f\u090f \u0939\u0948\u0902\u0964 \u0907\u0928\u094d\u0939\u0947\u0902 \u090f\u0915 \u0938\u093e\u0925 \u0938\u092e\u091d\u0928\u093e helpful \u0939\u0948, \u0915\u094d\u092f\u094b\u0902\u0915\u093f \u0939\u0930 model \u090f\u0915 \u0905\u0932\u0917 problem solve \u0915\u0930\u0924\u093e \u0939\u0948\u0964 \u0907\u0938 \u0916\u092c\u0930 \u092e\u0947\u0902 \u090f\u0915 \u0926\u093e\u0935\u093e \u092d\u0940 \u0939\u0948 \u091c\u093f\u0938\u0947 \u0927\u094d\u092f\u093e\u0928 \u0938\u0947 \u0938\u092e\u091d\u0928\u093e \u091a\u093e\u0939\u093f\u090f\u0964 CoRover \u0915\u093e \u0915\u0939\u0928\u093e \u0939\u0948 \u0915\u093f BharatGPT Mini \u0905\u092a\u0928\u0947 \u0938\u0947 15 \u0917\u0941\u0928\u093e \u092c\u0921\u093c\u0947 models \u0938\u0947 \u092c\u0947\u0939\u0924\u0930 perform \u0915\u0930\u0924\u093e \u0939\u0948\u0964 \u0932\u0947\u0915\u093f\u0928 reports \u092e\u0947\u0902 \u092f\u0939 \u092d\u0940 \u0938\u093e\u092b \u0932\u093f\u0916\u093e \u0917\u092f\u093e \u0939\u0948 \u0915\u093f company \u0928\u0947 \u0907\u0938 comparison \u0915\u0947 \u092a\u0940\u091b\u0947 \u0915\u094c\u0928 \u0938\u0947 benchmarks use \u0915\u093f\u090f, \u092f\u0939 specify \u0928\u0939\u0940\u0902 \u0915\u093f\u092f\u093e \u0939\u0948\u0964 AIKosh \u092a\u0930 listing \u0915\u093e \u092e\u0924\u0932\u092c \u092f\u0939 \u0928\u0939\u0940\u0902 \u0939\u0948 \u0915\u093f \u092f\u0939 models \u0924\u0941\u0930\u0902\u0924 \u0939\u0930 \u0915\u093f\u0938\u0940 \u0915\u0947 phone \u092e\u0947\u0902 \u0906 \u091c\u093e\u090f\u0902\u0917\u0947\u0964 \u0907\u0938\u0915\u093e \u092e\u0924\u0932\u092c \u0939\u0948 \u0915\u093f \u092f\u0939 models \u090f\u0915 national platform \u092a\u0930 available \u0939\u0941\u090f \u0939\u0948\u0902, \u091c\u0939\u093e\u0902 \u0938\u0947 developers, researchers \u0914\u0930 institutions \u0907\u0928 \u0924\u0915 \u092a\u0939\u0941\u0902\u091a \u0938\u0915\u0924\u0947 \u0939\u0948\u0902, \u0909\u0928\u094d\u0939\u0947\u0902 \u0938\u092e\u091d \u0938\u0915\u0924\u0947 \u0939\u0948\u0902 \u0914\u0930 \u0905\u092a\u0928\u0947 projects \u092e\u0947\u0902 use \u0915\u0930 \u0938\u0915\u0924\u0947 \u0939\u0948\u0902\u0964 \u0906\u092e user \u0915\u0947 \u0932\u093f\u090f \u0907\u0938 \u0916\u092c\u0930 \u0915\u093e \u0938\u0940\u0927\u093e \u0905\u0938\u0930 \u0924\u0941\u0930\u0902\u0924 \u0928\u0939\u0940\u0902 \u0926\u093f\u0916\u0947\u0917\u093e, \u0915\u094d\u092f\u094b\u0902\u0915\u093f \u092f\u0939 models mostly developers \u0914\u0930 companies \u0915\u0947 \u0932\u093f\u090f list \u0939\u0941\u090f \u0939\u0948\u0902, \u0928\u093e \u0915\u093f \u090f\u0915 consumer app \u0915\u0947 \u0930\u0942\u092a \u092e\u0947\u0902\u0964 \u0932\u0947\u0915\u093f\u0928 \u0932\u0902\u092c\u093e \u0905\u0938\u0930 \u0938\u092e\u091d\u0928\u093e \u0906\u0938\u093e\u0928 \u0939\u0948\u0964 CoRover \u0928\u0947 \u0905\u092a\u0928\u0947 3 BharatGPT models \u0915\u094b 28 October 2026 \u0915\u094b AIKosh platform \u092a\u0930 list \u0915\u093f\u092f\u093e\u0964<\/p>\n\n\n\n AIKosh \u0938\u0930\u0915\u093e\u0930 \u0915\u093e National AI Datasets and Models Platform \u0939\u0948, \u091c\u094b IndiaAI Mission \u0915\u0947 under \u091a\u0932\u093e\u092f\u093e \u091c\u093e\u0924\u093e \u0939\u0948, \u0914\u0930 \u091c\u0939\u093e\u0902 India \u092e\u0947\u0902 \u092c\u0928\u0947 datasets \u0914\u0930 models \u090f\u0915 \u091c\u0917\u0939 \u092a\u0930 \u092e\u093f\u0932\u0924\u0947 \u0939\u0948\u0902\u0964<\/p>\n\n\n\n BharatGPT Mini, \u091c\u094b 538 million parameters \u0935\u093e\u0932\u093e foundational model \u0939\u0948, BharatGPT Instruct, \u091c\u094b Mixture-of-Experts architecture \u0935\u093e\u0932\u093e multilingual model \u0939\u0948, \u0914\u0930 ASR Hindi Conformer, \u091c\u094b Hindi speech \u0915\u094b text \u092e\u0947\u0902 \u092c\u0926\u0932\u0924\u093e \u0939\u0948\u0964<\/p>\n\n\n\n Company \u0928\u0947 \u092f\u0939 \u0926\u093e\u0935\u093e \u0915\u093f\u092f\u093e \u0939\u0948, \u0932\u0947\u0915\u093f\u0928 \u0909\u0938\u0928\u0947 \u0907\u0938 comparison \u0915\u0947 benchmarks \u0915\u0947 \u0928\u093e\u092e specify \u0928\u0939\u0940\u0902 \u0915\u093f\u090f \u0939\u0948\u0902, \u0907\u0938\u0932\u093f\u090f \u0907\u0938 \u0926\u093e\u0935\u0947 \u0915\u094b verified fact \u0915\u0940 \u0924\u0930\u0939 \u0928\u0939\u0940\u0902 \u0932\u0947\u0928\u093e \u091a\u093e\u0939\u093f\u090f\u0964<\/p>\n\n\n\n \u0938\u0940\u0927\u093e \u092b\u093e\u092f\u0926\u093e \u0924\u0941\u0930\u0902\u0924 \u0928\u0939\u0940\u0902 \u0926\u093f\u0916\u0947\u0917\u093e, \u0932\u0947\u0915\u093f\u0928 \u0906\u0928\u0947 \u0935\u093e\u0932\u0947 time \u092e\u0947\u0902 Hindi \u0914\u0930 regional languages \u092e\u0947\u0902 \u092c\u0947\u0939\u0924\u0930 AI services, chatbots \u0914\u0930 voice assistants \u092e\u093f\u0932 \u0938\u0915\u0924\u0947 \u0939\u0948\u0902, \u0914\u0930 \u091b\u094b\u091f\u0947 models device \u092a\u0930 \u092d\u0940 \u091a\u0932 \u0938\u0915\u0924\u0947 \u0939\u0948\u0902\u0964<\/p>\n\n\n\n BharatGPT models \u0915\u093e AIKosh \u092a\u0930 \u0906\u0928\u093e India \u0915\u0947 AI ecosystem \u0915\u0947 \u0932\u093f\u090f \u090f\u0915 important step \u0939\u0948\u0964 \u092f\u0939 \u0926\u093f\u0916\u093e\u0924\u093e \u0939\u0948 \u0915\u093f India \u0938\u093f\u0930\u094d\u092b foreign models \u0915\u093e user \u0928\u0939\u0940\u0902 \u0930\u0939\u0928\u093e \u091a\u093e\u0939\u0924\u093e, \u092c\u0932\u094d\u0915\u093f \u0905\u092a\u0928\u0940 languages \u0915\u0947 \u0932\u093f\u090f \u0905\u092a\u0928\u0947 models \u092d\u0940 \u092c\u0928\u093e\u0928\u093e \u091a\u093e\u0939\u0924\u093e \u0939\u0948\u0964 Listing \u0938\u0947 reach \u092c\u0922\u093c\u0947\u0917\u0940, \u0932\u0947\u0915\u093f\u0928 \u0905\u0938\u0932\u0940 test \u0905\u092c \u0936\u0941\u0930\u0942 \u0939\u094b\u0917\u093e, \u091c\u092c developers \u0914\u0930 institutions \u0907\u0928 models \u0915\u094b \u0905\u092a\u0928\u0947 real projects \u092e\u0947\u0902 \u0906\u091c\u092e\u093e\u090f\u0902\u0917\u0947 \u0914\u0930 \u0909\u0928\u0915\u0940 performance \u0916\u0941\u0926 check \u0915\u0930\u0947\u0902\u0917\u0947\u0964 India \u092e\u0947\u0902 AI \u0915\u0940 \u092c\u093e\u0924 \u0905\u0915\u094d\u0938\u0930 \u092c\u0921\u093c\u0947 foreign models \u092a\u0930 \u0939\u094b\u0924\u0940 \u0939\u0948, \u091c\u094b mostly English \u092a\u0930 train \u0939\u0941\u090f \u0939\u0948\u0902\u0964 \u0932\u0947\u0915\u093f\u0928 28 October 2026 \u0915\u094b \u090f\u0915 \u0916\u092c\u0930 \u0906\u0908 \u091c\u094b India \u0915\u0947 \u0905\u092a\u0928\u0947 AI ecosystem \u0915\u0947 \u0932\u093f\u090f important \u0939\u0948\u0964 Gurugram \u0915\u0940 company CoRover.ai \u0928\u0947 \u0905\u092a\u0928\u0947 3 BharatGPT models \u0915\u094b AIKosh platform \u092a\u0930 list \u0915\u0930 \u0926\u093f\u092f\u093e \u0939\u0948\u0964AIKosh \u0915\u094d\u092f\u093e […]<\/p>\n","protected":false},"author":1,"featured_media":204,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"_jetpack_newsletter_access":"","_jetpack_dont_email_post_to_subs":false,"_jetpack_newsletter_tier_id":0,"_jetpack_memberships_contains_paywalled_content":false,"_jetpack_hide_featured_image":false,"_jetpack_featured_image_duplicate_dismissed":0,"_jetpack_feature_clip_id":0,"_jetpack_memberships_contains_paid_content":false,"footnotes":"","jetpack_post_was_ever_published":false},"categories":[1],"tags":[27,25,26,28],"class_list":["post-196","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-artificial-intelligence","tag-aikosh","tag-bharatgpt","tag-corover","tag-indiaai"],"jetpack_sharing_enabled":true,"jetpack_featured_media_url":"https:\/\/akshaymaurya.in\/blog\/wp-content\/uploads\/2026\/10\/thumb-bharatgpt.jpg","_links":{"self":[{"href":"https:\/\/akshaymaurya.in\/blog\/wp-json\/wp\/v2\/posts\/196","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/akshaymaurya.in\/blog\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/akshaymaurya.in\/blog\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/akshaymaurya.in\/blog\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/akshaymaurya.in\/blog\/wp-json\/wp\/v2\/comments?post=196"}],"version-history":[{"count":1,"href":"https:\/\/akshaymaurya.in\/blog\/wp-json\/wp\/v2\/posts\/196\/revisions"}],"predecessor-version":[{"id":197,"href":"https:\/\/akshaymaurya.in\/blog\/wp-json\/wp\/v2\/posts\/196\/revisions\/197"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/akshaymaurya.in\/blog\/wp-json\/wp\/v2\/media\/204"}],"wp:attachment":[{"href":"https:\/\/akshaymaurya.in\/blog\/wp-json\/wp\/v2\/media?parent=196"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/akshaymaurya.in\/blog\/wp-json\/wp\/v2\/categories?post=196"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/akshaymaurya.in\/blog\/wp-json\/wp\/v2\/tags?post=196"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}
BharatGPT \u0915\u094b \u092a\u0939\u0932\u0947 \u092d\u0940 India \u0915\u0947 \u092a\u0939\u0932\u0947 indigenous generative AI language model<\/a> \u0915\u0947 \u0930\u0942\u092a \u092e\u0947\u0902 \u092a\u0947\u0936 \u0915\u093f\u092f\u093e \u0917\u092f\u093e \u0925\u093e, \u091c\u094b 22 Indian languages \u092e\u0947\u0902 text \u0914\u0930 12 \u0938\u0947 \u091c\u094d\u092f\u093e\u0926\u093e languages \u092e\u0947\u0902 voice support \u0915\u093e \u0926\u093e\u0935\u093e \u0915\u0930\u0924\u093e \u0925\u093e\u0964 \u0932\u0947\u0915\u093f\u0928 \u0907\u0938 \u0928\u0908 \u0916\u092c\u0930 \u0915\u093e focus \u0935\u0939 \u092a\u0941\u0930\u093e\u0928\u093e \u0926\u093e\u0935\u093e \u0928\u0939\u0940\u0902 \u0939\u0948, \u092c\u0932\u094d\u0915\u093f \u092f\u0939 \u0939\u0948 \u0915\u093f \u0907\u0938 family \u0915\u0947 3 specific models \u0905\u092c AIKosh \u092a\u0930 list \u0939\u0941\u090f \u0939\u0948\u0902, \u091c\u093f\u0938\u0938\u0947 \u0909\u0928\u0915\u0940 reach \u090f\u0915 national platform \u0924\u0915 \u092c\u0922\u093c \u0917\u0908 \u0939\u0948\u0964
CoRover \u0915\u0947 founder \u0914\u0930 CEO Ankush Sabharwal \u0928\u0947 \u0907\u0938 move \u0915\u094b Indian developed AI capabilities \u0915\u094b national platform \u0915\u0947 through accessible \u092c\u0928\u093e\u0928\u0947 \u0915\u0940 \u0924\u0930\u092b \u090f\u0915 important \u0915\u0926\u092e \u092c\u0924\u093e\u092f\u093e \u0939\u0948\u0964 Simple \u092d\u093e\u0937\u093e \u092e\u0947\u0902, \u0905\u092c India \u092e\u0947\u0902 \u092c\u0928\u0947 \u092f\u0939 models \u0938\u093f\u0930\u094d\u092b \u090f\u0915 company \u0915\u0940 website \u0924\u0915 \u0938\u0940\u092e\u093f\u0924 \u0928\u0939\u0940\u0902 \u0930\u0939\u0947\u0902\u0917\u0947, \u092c\u0932\u094d\u0915\u093f \u090f\u0915 \u0938\u0930\u0915\u093e\u0930\u0940 platform \u092a\u0930 \u092d\u0940 \u0926\u093f\u0916\u0947\u0902\u0917\u0947, \u091c\u0939\u093e\u0902 \u0909\u0928\u094d\u0939\u0947\u0902 \u091c\u094d\u092f\u093e\u0926\u093e developers \u0914\u0930 institutions \u0926\u0947\u0916 \u0938\u0915\u0947\u0902\u0917\u0947\u0964<\/p>\n\n\n\n\u0915\u094c\u0928 \u0938\u0947 3 Models list \u0939\u0941\u090f \u0939\u0948\u0902<\/h2>\n\n\n\n
\u092a\u0939\u0932\u093e model \u0939\u0948 BharatGPT Mini\u0964 \u092f\u0939 \u090f\u0915 538 million parameters \u0935\u093e\u0932\u093e foundational language model \u0939\u0948, \u091c\u094b scratch \u0938\u0947 train \u0915\u093f\u092f\u093e \u0917\u092f\u093e \u0939\u0948\u0964 Scratch \u0938\u0947 train \u0915\u093e \u092e\u0924\u0932\u092c \u0939\u0948 \u0915\u093f \u0907\u0938\u0947 \u092a\u0939\u0932\u0947 \u0938\u0947 \u092c\u0928\u0947 \u0915\u093f\u0938\u0940 model \u092a\u0930 fine tune \u0915\u0930\u0915\u0947 \u0928\u0939\u0940\u0902, \u092c\u0932\u094d\u0915\u093f \u0936\u0941\u0930\u0942 \u0938\u0947 \u092c\u0928\u093e\u092f\u093e \u0917\u092f\u093e \u0939\u0948\u0964 \u0907\u0938\u0915\u093e size \u091b\u094b\u091f\u093e \u0939\u0948, \u0907\u0938\u0932\u093f\u090f \u0907\u0938\u0947 on-premises, cloud, edge \u092f\u093e hybrid setup \u092a\u0930 \u091a\u0932\u093e\u092f\u093e \u091c\u093e \u0938\u0915\u0924\u093e \u0939\u0948\u0964 On-premises \u0915\u093e \u092e\u0924\u0932\u092c \u0939\u0948 \u0905\u092a\u0928\u0947 \u0916\u0941\u0926 \u0915\u0947 server \u092a\u0930, edge \u0915\u093e \u092e\u0924\u0932\u092c \u0939\u0948 device \u0915\u0947 \u092a\u093e\u0938, \u091c\u0948\u0938\u0947 phone \u092f\u093e local machine \u092a\u0930, \u0914\u0930 hybrid \u0915\u093e \u092e\u0924\u0932\u092c \u0939\u0948 \u0926\u094b\u0928\u094b\u0902 \u0915\u093e combination\u0964
\u091b\u094b\u091f\u0947 model \u0915\u093e \u092b\u093e\u092f\u0926\u093e \u092f\u0939 \u0939\u0948 \u0915\u093f \u0907\u0938\u0947 \u091a\u0932\u093e\u0928\u0947 \u0915\u0947 \u0932\u093f\u090f \u092c\u0939\u0941\u0924 \u092c\u0921\u093c\u093e hardware \u0928\u0939\u0940\u0902 \u091a\u093e\u0939\u093f\u090f, \u0914\u0930 data \u0915\u094b device \u092a\u0930 \u0939\u0940 \u0930\u0916\u093e \u091c\u093e \u0938\u0915\u0924\u093e \u0939\u0948, \u091c\u094b privacy \u0915\u0947 \u0932\u093f\u090f \u0905\u091a\u094d\u091b\u093e \u0939\u094b\u0924\u093e \u0939\u0948\u0964 \u091b\u094b\u091f\u0940 companies \u0914\u0930 small businesses \u0915\u0947 \u0932\u093f\u090f \u092f\u0939 \u0916\u093e\u0938 \u0924\u094c\u0930 \u092a\u0930 useful \u0939\u094b \u0938\u0915\u0924\u093e \u0939\u0948, \u091c\u094b \u092c\u0921\u093c\u0947 models \u0915\u093e \u0916\u0930\u094d\u091a\u093e \u0928\u0939\u0940\u0902 \u0909\u0920\u093e \u0938\u0915\u0924\u0947\u0964
\u0926\u0942\u0938\u0930\u093e model \u0939\u0948 BharatGPT Instruct\u0964 \u092f\u0939 \u090f\u0915 multilingual model \u0939\u0948 \u091c\u094b Mixture-of-Experts architecture \u092a\u0930 \u092c\u0928\u093e \u0939\u0948\u0964 Mixture-of-Experts \u0915\u093e \u092e\u0924\u0932\u092c \u0939\u0948 \u0915\u093f model \u0915\u0947 \u0905\u0902\u0926\u0930 \u0915\u0908 \u091b\u094b\u091f\u0947 experts \u0939\u094b\u0924\u0947 \u0939\u0948\u0902, \u0914\u0930 \u0939\u0930 \u0915\u093e\u092e \u0915\u0947 \u0932\u093f\u090f \u0938\u093f\u0930\u094d\u092b \u091c\u0930\u0942\u0930\u0940 experts active \u0939\u094b\u0924\u0947 \u0939\u0948\u0902\u0964 \u0907\u0938\u0938\u0947 model \u092c\u0921\u093c\u093e \u0939\u094b\u0928\u0947 \u0915\u0947 \u092c\u093e\u0926 \u092d\u0940 efficient \u0930\u0939\u0924\u093e \u0939\u0948\u0964 \u092f\u0939 model text generation, coding \u0914\u0930 reasoning \u0915\u0947 \u0932\u093f\u090f design \u0915\u093f\u092f\u093e \u0917\u092f\u093e \u0939\u0948, \u0914\u0930 Indic \u0915\u0947 \u0938\u093e\u0925 global languages \u092e\u0947\u0902 \u092d\u0940 \u0915\u093e\u092e \u0915\u0930\u0924\u093e \u0939\u0948\u0964 \u0907\u0938\u0915\u093e \u0907\u0938\u094d\u0924\u0947\u092e\u093e\u0932 instruction following \u0914\u0930 conversational applications \u092e\u0947\u0902 \u0939\u094b\u0924\u093e \u0939\u0948, \u092e\u0924\u0932\u092c user \u0915\u0947 instruction \u092a\u0930 \u091c\u0935\u093e\u092c \u0926\u0947\u0928\u093e \u092f\u093e \u092c\u093e\u0924\u091a\u0940\u0924 \u0915\u0930\u0928\u093e\u0964
\u0924\u0940\u0938\u0930\u093e model \u0939\u0948 ASR Hindi Conformer\u0964 ASR \u0915\u093e \u092e\u0924\u0932\u092c \u0939\u0948 automatic speech recognition, \u092f\u093e\u0928\u0940 \u092c\u094b\u0932\u0940 \u0939\u0941\u0908 \u092c\u093e\u0924 \u0915\u094b text \u092e\u0947\u0902 \u092c\u0926\u0932\u0928\u093e\u0964 \u092f\u0939 model Hindi speech \u0915\u094b text \u092e\u0947\u0902 convert \u0915\u0930\u0924\u093e \u0939\u0948\u0964 Company \u0915\u0947 \u092e\u0941\u0924\u093e\u092c\u093f\u0915 AI4Bharat \u0914\u0930 Lahaja test set \u092a\u0930 \u0907\u0938\u0915\u093e word error rate 24.72 percent \u0914\u0930 character error rate 21.78 percent \u0939\u0948\u0964 Word error rate \u0915\u093e \u092e\u0924\u0932\u092c \u0939\u0948 \u0917\u0932\u0924 \u0936\u092c\u094d\u0926\u094b\u0902 \u0915\u093e percentage, \u0914\u0930 character error rate \u0915\u093e \u092e\u0924\u0932\u092c \u0939\u0948 \u0917\u0932\u0924 characters \u0915\u093e percentage\u0964 \u092f\u0939 numbers company \u0915\u0940 \u0924\u0930\u092b \u0938\u0947 \u0926\u093f\u090f \u0917\u090f test results \u0939\u0948\u0902, \u0914\u0930 \u0907\u0928\u0915\u093e \u092e\u0924\u0932\u092c \u092f\u0939 \u0939\u0948 \u0915\u093f Hindi speech \u0915\u094b text \u092e\u0947\u0902 \u092c\u0926\u0932\u0928\u0947 \u092e\u0947\u0902 \u0905\u092d\u0940 \u092d\u0940 \u0915\u093e\u092b\u0940 \u0917\u0932\u0924\u093f\u092f\u093e\u0902 \u0939\u094b \u0938\u0915\u0924\u0940 \u0939\u0948\u0902, \u0932\u0947\u0915\u093f\u0928 \u090f\u0915 Indian language \u0915\u0947 \u0932\u093f\u090f \u092f\u0939 \u090f\u0915 starting point \u0939\u0948 \u091c\u093f\u0938\u0947 \u0906\u0917\u0947 \u0938\u0941\u0927\u093e\u0930\u093e \u091c\u093e \u0938\u0915\u0924\u093e \u0939\u0948\u0964<\/p>\n\n\n\n15 \u0917\u0941\u0928\u093e \u0935\u093e\u0932\u093e \u0926\u093e\u0935\u093e \u0915\u094d\u092f\u093e \u0939\u0948<\/h2>\n\n\n\n
Benchmarks \u0915\u093e \u0928\u093e\u092e \u0928\u093e \u092c\u0924\u093e\u0928\u093e \u090f\u0915 \u092c\u0921\u093c\u093e caveat \u0939\u0948\u0964 \u091c\u092c \u0924\u0915 \u092f\u0939 \u092a\u0924\u093e \u0928\u093e \u0939\u094b \u0915\u093f \u0915\u093f\u0938 test \u092a\u0930, \u0915\u093f\u0938 data \u092a\u0930 \u0914\u0930 \u0915\u093f\u0938 condition \u092e\u0947\u0902 comparison \u0939\u0941\u0906, \u0924\u092c \u0924\u0915 \u0907\u0938 \u0926\u093e\u0935\u0947 \u0915\u094b fact \u0915\u0940 \u0924\u0930\u0939 \u0928\u0939\u0940\u0902 \u0932\u0947\u0928\u093e \u091a\u093e\u0939\u093f\u090f\u0964 \u0907\u0938\u0932\u093f\u090f \u0907\u0938 article \u092e\u0947\u0902 \u0907\u0938 \u0926\u093e\u0935\u0947 \u0915\u094b \u0938\u093f\u0930\u094d\u092b company \u0915\u093e \u0926\u093e\u0935\u093e \u092c\u0924\u093e\u092f\u093e \u091c\u093e \u0930\u0939\u093e \u0939\u0948, \u0928\u093e \u0915\u093f \u090f\u0915 verified result\u0964
\u0907\u0938\u0940 \u0924\u0930\u0939 BharatGPT Instruct \u0915\u0947 \u092c\u093e\u0930\u0947 \u092e\u0947\u0902 \u092d\u0940 \u092a\u0939\u0932\u0947 \u0915\u0941\u091b benchmark results \u0906\u090f \u0925\u0947, \u091c\u093f\u0928\u092e\u0947\u0902 company \u0928\u0947 \u0915\u0939\u093e \u0925\u093e \u0915\u093f \u092f\u0939 \u0915\u0941\u091b reasoning tests \u092a\u0930 \u092c\u0921\u093c\u0947 models \u0938\u0947 \u0906\u0917\u0947 \u0930\u0939\u093e\u0964 \u0935\u0939 results \u092d\u0940 company \u0915\u0940 \u0924\u0930\u092b \u0938\u0947 \u0925\u0947, \u0907\u0938\u0932\u093f\u090f \u0909\u0928\u094d\u0939\u0947\u0902 \u092d\u0940 \u0909\u0938\u0940 caution \u0915\u0947 \u0938\u093e\u0925 \u0926\u0947\u0916\u0928\u093e \u091a\u093e\u0939\u093f\u090f\u0964 AI \u092e\u0947\u0902 benchmark \u0926\u093e\u0935\u0947 \u0905\u0915\u094d\u0938\u0930 marketing \u0915\u093e \u0939\u093f\u0938\u094d\u0938\u093e \u0939\u094b\u0924\u0947 \u0939\u0948\u0902, \u0914\u0930 \u0905\u0938\u0932\u0940 performance \u0924\u092c \u092a\u0924\u093e \u091a\u0932\u0924\u0940 \u0939\u0948 \u091c\u092c developers \u0907\u0928 models \u0915\u094b \u0905\u092a\u0928\u0947 real \u0915\u093e\u092e \u092e\u0947\u0902 \u0906\u091c\u092e\u093e\u0924\u0947 \u0939\u0948\u0902\u0964<\/p>\n\n\n\nAIKosh \u092a\u0930 listing \u0915\u093e \u092e\u0924\u0932\u092c \u0915\u094d\u092f\u093e \u0939\u0948<\/h2>\n\n\n\n
India \u0915\u0947 \u0932\u093f\u090f \u0907\u0938\u0915\u093e \u092c\u0921\u093c\u093e \u092e\u0924\u0932\u092c \u0939\u0948 sovereign AI \u0915\u0940 direction \u092e\u0947\u0902 \u090f\u0915 \u0915\u0926\u092e\u0964 Sovereign AI \u0915\u093e \u092e\u0924\u0932\u092c \u0939\u0948 \u0910\u0938\u093e AI \u091c\u094b \u0915\u093f\u0938\u0940 \u0926\u0947\u0936 \u0915\u0940 \u0905\u092a\u0928\u0940 languages, data \u0914\u0930 \u091c\u0930\u0942\u0930\u0924\u094b\u0902 \u0915\u0947 \u0932\u093f\u090f \u092c\u0928\u093e \u0939\u094b, \u0914\u0930 \u091c\u093f\u0938\u0915\u093e control \u0909\u0938 \u0926\u0947\u0936 \u0915\u0947 \u0939\u093e\u0925 \u092e\u0947\u0902 \u0939\u094b\u0964 \u091c\u092c models foreign companies \u092a\u0930 depend \u0915\u0930\u0924\u0947 \u0939\u0948\u0902, \u0924\u094b data, price \u0914\u0930 availability \u0909\u0928\u0915\u0940 policies \u092a\u0930 depend \u0915\u0930\u0924\u0940 \u0939\u0948\u0964 \u0905\u092a\u0928\u0947 models \u0939\u094b\u0928\u0947 \u0938\u0947 India \u0915\u0940 companies \u0914\u0930 \u0938\u0930\u0915\u093e\u0930\u0940 institutions \u0915\u094b \u090f\u0915 alternative \u092e\u093f\u0932\u0924\u093e \u0939\u0948\u0964
\u092f\u0939 listing \u0916\u093e\u0938 \u0924\u094c\u0930 \u092a\u0930 \u0909\u0928 sectors \u0915\u0947 \u0932\u093f\u090f useful \u0939\u094b \u0938\u0915\u0924\u0940 \u0939\u0948 \u091c\u0939\u093e\u0902 Hindi \u0914\u0930 regional languages \u092e\u0947\u0902 \u0915\u093e\u092e \u0939\u094b\u0924\u093e \u0939\u0948, \u091c\u0948\u0938\u0947 \u0938\u0930\u0915\u093e\u0930\u0940 services, banking, education \u0914\u0930 customer support\u0964 \u0905\u0917\u0930 \u090f\u0915 model Hindi speech \u0915\u094b \u0905\u091a\u094d\u091b\u0940 \u0924\u0930\u0939 \u0938\u092e\u091d \u0938\u0915\u0947, \u0924\u094b \u0935\u0939 \u0932\u094b\u0917 \u092d\u0940 AI \u0915\u093e \u092b\u093e\u092f\u0926\u093e \u0909\u0920\u093e \u0938\u0915\u0924\u0947 \u0939\u0948\u0902 \u091c\u094b English \u092e\u0947\u0902 comfortable \u0928\u0939\u0940\u0902 \u0939\u0948\u0902\u0964<\/p>\n\n\n\n\u0906\u092e User \u0915\u0947 \u0932\u093f\u090f \u0907\u0938\u0915\u093e \u0915\u094d\u092f\u093e \u092e\u0924\u0932\u092c \u0939\u0948<\/h2>\n\n\n\n
\u0905\u0917\u0930 Indian languages \u0915\u0947 models \u092c\u0947\u0939\u0924\u0930 \u0939\u094b\u0902\u0917\u0947, \u0924\u094b \u0906\u0928\u0947 \u0935\u093e\u0932\u0947 time \u092e\u0947\u0902 \u0906\u092a\u0915\u094b Hindi \u0914\u0930 \u0905\u092a\u0928\u0940 regional language \u092e\u0947\u0902 \u092c\u0947\u0939\u0924\u0930 chatbots, voice assistants \u0914\u0930 \u0938\u0930\u0915\u093e\u0930\u0940 services \u092e\u093f\u0932 \u0938\u0915\u0924\u0940 \u0939\u0948\u0902\u0964 \u091b\u094b\u091f\u0947 models \u091c\u094b device \u092a\u0930 \u091a\u0932 \u0938\u0915\u0924\u0947 \u0939\u0948\u0902, \u0935\u0939 internet \u0928\u093e \u0939\u094b\u0928\u0947 \u092a\u0930 \u092d\u0940 \u0915\u093e\u092e \u0915\u0930 \u0938\u0915\u0924\u0947 \u0939\u0948\u0902, \u0914\u0930 data \u0915\u094b device \u092a\u0930 \u0939\u0940 \u0930\u0916 \u0938\u0915\u0924\u0947 \u0939\u0948\u0902, \u091c\u094b privacy \u0915\u0947 \u0932\u093f\u090f \u0905\u091a\u094d\u091b\u093e \u0939\u0948\u0964
\u0926\u0942\u0938\u0930\u0940 \u0924\u0930\u092b, \u092f\u0939 \u092d\u0940 \u0938\u091a \u0939\u0948 \u0915\u093f \u0907\u0928 models \u0915\u0940 \u0905\u0938\u0932\u0940 quality \u0924\u092d\u0940 \u092a\u0924\u093e \u091a\u0932\u0947\u0917\u0940 \u091c\u092c \u0932\u094b\u0917 \u0907\u0928\u094d\u0939\u0947\u0902 real \u0915\u093e\u092e \u092e\u0947\u0902 use \u0915\u0930\u0947\u0902\u0917\u0947\u0964 Listing \u090f\u0915 \u0936\u0941\u0930\u0941\u0906\u0924 \u0939\u0948, final \u092e\u0902\u091c\u093f\u0932 \u0928\u0939\u0940\u0902\u0964 \u0907\u0938\u0932\u093f\u090f \u0907\u0938 \u0916\u092c\u0930 \u0915\u094b \u090f\u0915 positive step \u0915\u0940 \u0924\u0930\u0939 \u0926\u0947\u0916\u0928\u093e \u091a\u093e\u0939\u093f\u090f, \u0932\u0947\u0915\u093f\u0928 \u0909\u0938\u0915\u0947 \u0938\u093e\u0925 \u092f\u0939 \u092d\u0940 \u092f\u093e\u0926 \u0930\u0916\u0928\u093e \u091a\u093e\u0939\u093f\u090f \u0915\u093f benchmark \u0926\u093e\u0935\u0947 \u0905\u092d\u0940 verify \u0939\u094b\u0928\u0947 \u092c\u093e\u0915\u0940 \u0939\u0948\u0902\u0964<\/p>\n\n\n\nFAQ<\/h2>\n\n\n\n
BharatGPT models AIKosh \u092a\u0930 \u0915\u092c list \u0939\u0941\u090f?<\/h3>\n\n\n\n
AIKosh \u0915\u094d\u092f\u093e \u0939\u0948?<\/h3>\n\n\n\n
\u0915\u094c\u0928 \u0938\u0947 3 models list \u0939\u0941\u090f \u0939\u0948\u0902?<\/h3>\n\n\n\n
\u0915\u094d\u092f\u093e BharatGPT Mini \u0938\u091a \u092e\u0947\u0902 15 \u0917\u0941\u0928\u093e \u092c\u0921\u093c\u0947 models \u0938\u0947 \u092c\u0947\u0939\u0924\u0930 \u0939\u0948?<\/h3>\n\n\n\n
\u0906\u092e user \u0915\u094b \u0907\u0938\u0938\u0947 \u0915\u094d\u092f\u093e \u092b\u093e\u092f\u0926\u093e \u0939\u094b\u0917\u093e?<\/h3>\n\n\n\n
\u0906\u0916\u093f\u0930\u0940 \u092c\u093e\u0924<\/h2>\n\n\n\n
External source: AI Daily Post \u0915\u0940 report<\/a><\/p>\n","protected":false},"excerpt":{"rendered":"