Sarvam AI
What its model cards measure
Tall ticks are Sarvam AI model cards that list the benchmark in their headline table. Gray dots are cards that leave it out. Rows are the benchmarks the lab listed most, at least twice each, in the order it started using them.
Model cards
| Date | Model | Document | Benchmarks |
|---|---|---|---|
| 6 Mar 2026 | Sarvam 30B | Open-sourcing Sarvam 30B & 105B models (Sarvam AI blog) | 16 |
| 6 Mar 2026 | Sarvam 105B | Open-sourcing Sarvam 30B & 105B models (Sarvam AI blog) | 17 |
| 23 May 2025 | Sarvam-M | Sarvam-M: Explorations in Post Training and Inferencing Optimizations for a Hybrid Indic LLM (Sarvam AI blog) | 12 |
Every Sarvam AI model
- Sarvam 105B Mar 2026
- Sarvam 30B Mar 2026
- Sarvam-M May 2025