Esc to close · ⌘K / Ctrl-K opens search anywhere
Not sure which model? Pick by use case → ranked by measured ₹/task, first-token latency and throughput.
| Model | Routes | ₹ / Mtok (in · out) ↕ | Languages | Latency | Status |
|---|---|---|---|---|---|
| Loading live catalog… | |||||
Pricing is indicative for the developer beta. Residency-tagged routes are eligible for
data_policy: "india_only". Data refreshes on page load from
/v1/models.
Routes marked (off) work today with
bring-your-own-key: pass your provider key as
upstream_key and BharatRouter does the piping.
Curated from Hugging Face. Want one of these served on BharatRouter? Email [email protected] — open-weight models can be hosted on demand. Explore more: Indic models on Hugging Face ↗ · AIKosh — India's national AI repository ↗
| Model | Org | Size | Languages | Access |
|---|---|---|---|---|
| Krutrim 2 Instruct Open weights on HF — not currently served via the BharatRouter API | Krutrim (Ola) | 12B | 22 Indian + English | open weights |
| Sarvam 105B Flagship; available via BYOK | Sarvam AI | 105B | 22 Indian + English | API · BYOKopen weights |
| Sarvam 30B Production workhorse; available via BYOK | Sarvam AI | 30B | 10 Indian + English | API · BYOKopen weights |
| Param 2 Thinking Public-good sovereign model; hosting on request | BharatGen (Govt-backed) | 17B MoE (2.4B active) | 22 Indian + English | open weights |
| Param 1 Instruct Fits small GPUs; hosting on request | BharatGen | 2.9B | Hindi + English | open weights |
| Sarvam-M Open weights (API retired); hosting on request | Sarvam AI | 24B | 11 Indian + English | open weights |
| Fathom R1 Indian open reasoning model; hosting on request | Fractal AI Research | 14B | English (reasoning) | open weights |
| Airavata Hindi instruct; hosting on request | AI4Bharat (IIT Madras) | 7B | Hindi + English | open weights |
| OpenHathi Early Hindi base model | Sarvam AI | 7B | Hindi + English | open weights |
| Navarasa 2.0 Community Indic fine-tune | Telugu LLM Labs | 7B | 15 Indian + English | open weights |
| Tamil-LLaMA Tamil instruct family (also 13B, Malayalam) | Community (abhinand) | 7B | Tamil + English | open weights |
| AyurParam / AgriParam / FinanceParam Vertical sovereign models (ayurveda, agriculture, finance) | BharatGen | 2.9B each | Hindi + English | open weights |
| Pragna Edge-friendly Indic model | Soket AI Labs | 1B | Hindi, Bangla, Gujarati + English | open weights |
| Ganga Academic Indic model | Lingo, IIT Gandhinagar | 1B | Hindi + English | open weights |