Bengaluru-based Sarvam AI has reported strong benchmark results for its Sarvam Vision model, claiming superior performance on document understanding, OCR and multi-script Indian-language tasks. On olmOCR-Bench it reported 84.3% accuracy versus Google Gemini 3 Pro's 80.20 and OpenAI GPT 5.2's 69.80; the company will demonstrate its models ahead of the India AI Impact Summit on 16 February in New Delhi.
Key Points
- 1Achieves 84.3% on olmOCR-Bench, outperforming Gemini 3 Pro 80.20 and GPT 5.2 69.80
- 2Highlights domain-specific training improves handling of non-Latin scripts and complex page layouts
- 3Suggests practitioners should prioritize task-specific models and Indic datasets for better document parsing
Scoring Rationale
Notable benchmark gains on Indic document tasks, but results rely on company-reported tests lacking independent validation.
Sources
Public references used for this report.
Practice interview problems based on real data
1,625 SQL & Python problems across 15 industry datasets — the exact type of data you work with.
Try 250 free problems
