Sarvam 30B (High) — benchmark results
Sarvam 30B evaluated at the high reasoning-effort setting. Provider: Sarvam AI. Released 2026-02-18. Access: API.
Unified ELO 1489 ± 27, rank #844 of 1776 rated models, from 33 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| CritPt | 0.3 | Accuracy (self-reported) | 65 |
| AA CritPt | 0.29 | Accuracy (%) | 58.7 |
| AA Humanity's Last Exam | 7.04 | Accuracy (%) | 50.8 |
| AA TAU-2 Bench | 34.5 | Accuracy (%) | 42.4 |
| AA GPQA Diamond | 63.33 | Accuracy (%) | 42.2 |
| AA Omniscience - Software Engineering (SWE) - Go | 14 | Accuracy (%) | 39.6 |
| AA Omniscience - Software Engineering (SWE) - Julia | 8 | Accuracy (%) | 37.3 |
| AA Omniscience - Science, Engineering & Mathematics | 21 | Accuracy (%) | 32.4 |
| AA Omniscience - Software Engineering (SWE) - Rust | 42 | Accuracy (%) | 32.4 |
| AA SciCode | 19.21 | Accuracy (%) | 20.5 |
| AA Omniscience - Business | 11.4 | Accuracy (%) | 20.3 |
| AA Omniscience - Law | 7.5 | Accuracy (%) | 20.3 |
Interactive version: theaggregate.ai/model?slug=sarvam-30b-high · How the rankings work · Data refreshed daily, snapshot 2026-07-22.