BharatGen Text Models
BharatGen’s text models are large language models built for Indian languages, covering general-purpose text generation as well as domain-specific applications across healthcare, agriculture, finance and law.
Param2 is BharatGen’s most advanced text model: a 17B Mixture-of-Experts architecture that activates 2.4B parameters per token. It handles reasoning, chain-of-thought, tool calling, mathematics and code generation, and is trained on approximately 22 trillion tokens.
Supported languages: English + all 22 Scheduled Indian languages, namely Hindi, Assamese, Bengali, Bodo, Dogri, Gujarati, Kannada, Konkani, Kashmiri, Maithili, Malayalam, Manipuri, Marathi, Nepali, Oriya, Punjabi, Sanskrit, Santali, Sindhi, Tamil, Telugu and Urdu.
Param1 is BharatGen’s bilingual foundation model family, pre-trained and built from scratch in English and Hindi. BharatGen has also developed four domain-specific models: AyurParam for Ayurveda and classical medical texts, AgriParam for agriculture, FinanceParam for financial queries and LegalParam for India-centric law across 14 legal domains.
No. |
Text Models Title |
Try on Hugging Face |
View on AI Kosh |
|---|---|---|---|
1. |
BharatGen Param2 17B |  |
|
2. |
BharatGen Param1 7B |
|
|
3. |
BharatGen Param1 2.9B Instruct |
| |
4. |
BharatGen Param1 |
|
|
5. |
BharatGen AyurParam |
| |
6. |
BharatGen AgriParam |
| |
7. |
BharatGen FinanceParam |
| |
8. |
BharatGen LegalParam |
| |
No. |
BharatGen Text Models Title |
View Details |
|---|---|---|
1. |
BharatGen - Param 1: Indic-Scale Bilingual Foundation Model - aikosh | |
2. |
BharatGen - Patram 7B Instruct - aikosh | |
3. |
Patram-7b-instruct | |
4. |
Param1-2.9B-Instruct | |
5. |
Param2-17B-A2.4B-Thinking | |
6. |
AyurParam | |
7. |
AgriParam | |
8. |
FinanceParam | |
9. |
Bhashabench-Ayur | |
10. |
Bhashabench-Krishi | |
11. |
Bhashabench-Finance | |