Neura Intelligence Index
Category Comparison
Benchmark-by-Benchmark
reasoning
| Benchmark | Claude 3.5 Sonnet | Sarvam-30B |
|---|---|---|
| GPQA Diamond | 67.2 | 66.5 |
coding
| Benchmark | Claude 3.5 Sonnet | Sarvam-30B |
|---|---|---|
| SWE-bench Verified | 49 | 34 |
multimodal
| Benchmark | Claude 3.5 Sonnet | Sarvam-30B |
|---|---|---|
| MMMU | 68.3 | — |
agent
| Benchmark | Claude 3.5 Sonnet | Sarvam-30B |
|---|---|---|
| TAU-Bench Retail | 69.2 | — |
| BrowseComp | — | 35.5 |
math
| Benchmark | Claude 3.5 Sonnet | Sarvam-30B |
|---|---|---|
| AIME 2025 | — | 96.7 |
Category Winners
coding
15.8
math
S
Sarvam-30B80.1
reasoning
49.1
multimodal
51.8
agent
54.3