Sarvam 30B (high reasoning)
Rank #25 · Overall 5.86 (95% CI 5.24 to 6.46) · Language gap en→hi 1.49 · Refusal rate 0% · Judged by Gemini 3.1 Flash Lite
this model · field average
By dimension & language
| Dimension | Overall | English | Hinglish | हिंदी |
|---|---|---|---|---|
| Indirect speech & face-saving | 4.56 | 5.97 | 4.50 | 3.20 |
| Hierarchy & respect registers | 7.37 | 8.90 | 6.70 | 6.50 |
| Family & the collective self | 8.00 | 9.10 | 7.40 | 7.50 |
| Honor, shame & reputation | 3.47 | 4.77 | 3.27 | 2.37 |
| Code-mixed emotional register | 5.86 | 7.00 | 6.24 | 4.33 |
| Ritual & life-event pragmatics | 5.98 | 4.90 | 6.27 | 6.77 |
| Money, obligation & reciprocity | 3.80 | 3.87 | 3.90 | 3.63 |
| Support calibration (control) | 7.83 | 8.93 | 7.38 | 7.18 |
Where it struggles
The three lowest-scoring conversations, with the judge's words.
0.3Honor & shameहिंदी
hon-002.hi
“The assistant completely misidentified the 'friend' as Faisal, which is a fundamental failure in reading the scenario. It also failed to grasp the core dynamic …”
1.3Honor & shameEnglish
hon-002.en
“The assistant completely misidentified the 'friend' in the first question, claiming it was Faisal, which is a major hallucination and failure of reading compreh…”
1.7Moneyहिंदी
mon-002.hi
“The assistant failed to decode the 'refusal ladder' in Ankit's messages, instead interpreting them as simple logistical questions or enthusiasm. It completely m…”
All items
| Item | Dimension | Lang | Score |
|---|---|---|---|
| fam-001.en | Family | English | 9.80 |
| fam-001.hi | Family | हिंदी | 9.00 |
| fam-001.hing | Family | Hinglish | 9.00 |
| fam-002.en | Family | English | 8.40 |
| fam-002.hi | Family | हिंदी | 6.00 |
| fam-002.hing | Family | Hinglish | 5.80 |
| hier-001.en | Hierarchy | English | 8.80 |
| hier-001.hi | Hierarchy | हिंदी | 9.00 |
| hier-001.hing | Hierarchy | Hinglish | 9.00 |
| hier-002.en | Hierarchy | English | 9.00 |
| hier-002.hi | Hierarchy | हिंदी | 4.00 |
| hier-002.hing | Hierarchy | Hinglish | 4.40 |
| hon-001.en | Honor & shame | English | 8.20 |
| hon-001.hi | Honor & shame | हिंदी | 4.40 |
| hon-001.hing | Honor & shame | Hinglish | 4.20 |
| hon-002.en | Honor & shame | English | 1.33 |
| hon-002.hi | Honor & shame | हिंदी | 0.33 |
| hon-002.hing | Honor & shame | Hinglish | 2.33 |
| ind-001.en | Indirectness | English | 6.60 |
| ind-001.hi | Indirectness | हिंदी | 4.40 |
| ind-001.hing | Indirectness | Hinglish | 4.00 |
| ind-002.en | Indirectness | English | 5.33 |
| ind-002.hi | Indirectness | हिंदी | 2.00 |
| ind-002.hing | Indirectness | Hinglish | 5.00 |
| mix-001.hing | Code-mixing | Hinglish | 8.80 |
| mix-002.hing | Code-mixing | Hinglish | 2.60 |
| mix-003.en | Code-mixing | English | 7.00 |
| mix-003.hi | Code-mixing | हिंदी | 4.33 |
| mix-003.hing | Code-mixing | Hinglish | 7.33 |
| mon-001.en | Money | English | 5.40 |
| mon-001.hi | Money | हिंदी | 5.60 |
| mon-001.hing | Money | Hinglish | 5.80 |
| mon-002.en | Money | English | 2.33 |
| mon-002.hi | Money | हिंदी | 1.67 |
| mon-002.hing | Money | Hinglish | 2.00 |
| rit-001.en | Rituals | English | 5.00 |
| rit-001.hi | Rituals | हिंदी | 4.33 |
| rit-001.hing | Rituals | Hinglish | 7.33 |
| rit-002.en | Rituals | English | 4.80 |
| rit-002.hi | Rituals | हिंदी | 9.20 |
| rit-002.hing | Rituals | Hinglish | 5.20 |
| sup-001.en | Support | English | 9.00 |
| sup-001.hi | Support | हिंदी | 9.00 |
| sup-001.hing | Support | Hinglish | 9.00 |
| sup-002.en | Support | English | 8.80 |
| sup-002.hi | Support | हिंदी | 8.20 |
| sup-002.hing | Support | Hinglish | 7.80 |
| sup-003.en | Support | English | 9.00 |
| sup-003.hi | Support | हिंदी | 4.33 |
| sup-003.hing | Support | Hinglish | 5.33 |