HELM SeaHELM - TyDiQA — leaderboard
Metric: SQuAD macro-averaged F1 score. Source: crfm.stanford.edu. 21 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 2.0 Flash | 81.51 |
| 2 | Gemini 2.0 Flash Lite | 65.01 |
| 3 | Llama 4 Maverick Instruct FP8 | 64.9 |
| 4 | GPT-4o (2024-11-20) | 63.68 |
| 5 | Gemma 2 9B (IT) | 62.17 |
| 6 | Llama 3.2 90B Vision Instruct | 60.07 |
| 7 | Llama 3.3 70B Instruct | 57.34 |
| 8 | Gemma 2 27B (IT) | 56.32 |
| 9 | Llama 4 Scout Instruct | 55.91 |
| 10 | GPT-4o Mini (2024-07-18) | 52.46 |
| 11 | Llama 3.1 405B Instruct | 52.2 |
| 12 | Llama 3.1 70B Instruct | 51.04 |
| 13 | Qwen 2 72B Instruct | 50.52 |
| 14 | DeepSeek R1 Distill Llama 8B | 50.19 |
| 15 | Qwen 2.5 72B Instruct | 40.7 |
Interactive version: theaggregate.ai/benchmark?slug=helm-seahelm-tydiqa · How the rankings work · Data refreshed daily, snapshot 2026-07-22.