SA
CompareSarvam 105B
sarvam/sarvam-105b
Flagship Indian-language reasoning model for enterprise multilingual applications
- Context window
- 131K
- Input
- Text
- Output
- Text
- Released
- 2025-09-01
- Max output
- 131K
- Model parameters
- 106B
- Weights
- Open
- License
- apache-2.0
- Capabilities
- Open weight, Reasoning, Temperature, Tool call
IN
INR4.00
INR / 1,000,000 token
OUT
INR16.00
INR / 1,000,000 token
Cache read
INR2.50
INR / 1,000,000 token
Performance
Overall capability
36.7/100
- Coding
- —
- Reasoning
- 38.4
- Professional work
- —
- Knowledge & language
- 14.0
| Humanity’s Last ExamSarvam 105B (High) | Reasoning | Artificial Analysis | 2026-11-05 | |
| AA-LCRSarvam 105B (High) | Knowledge & language | Artificial Analysis | 2026-11-05 | |
| CritPtSarvam 105B (High) | Reasoning | Artificial Analysis | 2026-11-05 | |
| IFBench（Artificial Analysis）Sarvam 105B (High) | Knowledge & language | Artificial Analysis | 2026-11-05 | |
| AA-OmniscienceSarvam 105B (High) | Knowledge & language | Artificial Analysis | 2026-10-27 |
Providers
| Input | Output | Cache read | Cache write | ||||
|---|---|---|---|---|---|---|---|
| FastRouter | sarvam/sarvam-105b | Openai chat | - | - | - | ||
| NanoGPT | sarvam-105b | Openai chat | - | - | |||
| Sarvam AI | sarvam-105b | Custom | - | - |
ModelRegistry · Catalog updated 2026-11-05