HELM SeaHELM - IndicQA — leaderboard

Metric: SQuAD macro-averaged F1 score. Source: crfm.stanford.edu. 21 models tracked.

Top models

#ModelScore
1Llama 3.2 90B Vision Instruct57.23
2Llama 4 Maverick Instruct FP848.42
3Gemma 2 27B (IT)47.24
4Llama 3.1 70B Instruct43.31
5Llama 3.3 70B Instruct39.05
6GPT-4o (2024-11-20)35.44
7Llama 4 Scout Instruct34
8Llama 3.1 405B Instruct32.7
9Gemini 2.0 Flash Lite31.54
10Gemini 2.0 Flash29.76
11GPT-4o Mini (2024-07-18)29.39
12Qwen 2 72B Instruct27.21
13Gemma 2 9B (IT)26.25
14DeepSeek V322.69
15Qwen 2.5 72B Instruct19.16

Interactive version: theaggregate.ai/benchmark?slug=helm-seahelm-indicqa · How the rankings work · Data refreshed daily, snapshot 2026-07-22.