Sarvam AI model product
Sarvam-105B is Sarvam AI's flagship open-source Mixture-of-Experts reasoning model built for complex reasoning, coding, and agentic workflows. It uses 128 sparse experts with Multi-head Latent Attention for efficient long-context inference and was pre-trained on 12 trillion tokens spanning code, mathematics, multilingual, and web data.
Updated Aug 10, 2026. Default version: Sarvam-105B
Structured fields from the published default version.
This profile uses the latest version under this unique model that has a calculated LLMBoard score. Arena and price are excluded.
Published benchmark records for the scored version Sarvam-105B.
| MATH-500 | 1.0 | 2 | 32 | 96.8% | C | |
| Beyond AIME | 0.7 | 3 | 5 | 50.0% | C | |
| MMLU | 0.9 | 6 | 100 | 95.0% | C | |
| Arena-Hard v2 | 0.7 | 7 | 16 | 60.0% | C | |
| HMMT25 | 0.9 | 10 | 25 | 62.5% | C | |
| AIME 2025 | 1.0 | 17 | 114 | 85.8% | C | |
| HMMT 2025 | 0.9 | 21 | 33 | 37.5% | C | |
| LiveCodeBench v6 | 0.7 | 27 | 53 | 50.0% | C | |
| IFEval | 0.8 | 39 | 65 | 40.6% | C | |
| MMLU-Pro | 0.8 | 41 | 129 | 68.8% | C | |
| BrowseComp | 0.5 | 45 | 58 | 22.8% | C | |
| Humanity's Last Exam | 0.1 | 78 | 92 | 15.4% | C | |
| SWE-Bench Verified | 0.5 | 90 | 104 | 13.6% | C | |
| GPQA | 0.8 | 93 | 233 | 60.3% | C |
Preference and agent-evaluation signals from published Arena datasets.
The default version has no published Arena rows, or its source alias has not been resolved.
Official vendor API PAYG pricing is summarized first. The table then lists individual provider offerings without treating their minimum as the official price.
| NanoGPT | sarvam-105b | global | $0.045 | $0.177 | 131.1K |
Official prices use only the vendor's configured official Provider and positive standard USD PAYG rates. Third-party offers remain explicitly labeled.
All published versions linked to this unique model. The score columns identify the version used by the current overall ranking.
| Sarvam-105B | 52.8 | 105B | 131.1K | 131.1K | Yes | Apache 2.0 |
A concise description based on the published model registry.
Sarvam-105B is Sarvam AI's flagship open-source Mixture-of-Experts reasoning model built for complex reasoning, coding, and agentic workflows. It uses 128 sparse experts with Multi-head Latent Attention for efficient long-context inference and was pre-trained on 12 trillion tokens spanning code, mathematics, multilingual, and web data.
Use the benchmark, Arena and pricing sections above as separate evidence. A missing field means the current data snapshot does not support that claim.
Data snapshot: 2026-08-07. Editorial model content is not available in the backend.
Open a comparison with the three ranked models immediately above and below this model.
Recommendations prioritize the same model type and family, then the closest published LLMBoard score.
Common questions about Sarvam 105B.
Sarvam 105B's default version was released on Mar 6, 2026.
No official standard PAYG price is currently available for Sarvam 105B. The lowest tracked third-party offer starts at $0.045 input and $0.177 output via NanoGPT.
Sarvam 105B is published under Sarvam AI in the model registry.
The default version has a 131.1K token context window.
Yes. The default version is marked as open weight under Apache 2.0.
1 published provider offerings are linked to the default version.
Nearby ranked alternatives include Qwen3.5 35B A3B, Qwen2.5 72B, Gemini 2.0 Flash.