HellaSwag: scores, leaders and methodology
HellaSwag leaderboard data from the index. Commonsense sentence completion with adversarial wrong answers. The top score is 96.1 and the average is 81.4.
What the data shows
GPT-5.2 Max leads at 96.1. The gap to the average is 14.7 points, and the gap to the 10th-ranked model is 1.7 points.
The median score is 83.6. Models above that line are in the upper half of the index for this benchmark.
How to read the ranking
- Use the score to shortlist, not to declare a winner.
- Check price and context for the top candidates.
- Read the benchmark methodology before quoting a score.
- Pair this benchmark with a task-specific evaluation.
Top 20 on HellaSwag
| # | Model | Provider | Score | Input / 1M | Context |
|---|---|---|---|---|---|
| 1 | GPT-5.2 Max | OpenAI | 96.1 | $21.39 | 1,000,000 |
| 2 | GPT-5.2 Turbo | OpenAI | 95.5 | $20.73 | 1,000,000 |
| 3 | GPT-5.2 Pro | OpenAI | 95.3 | $21.32 | 2,000,000 |
| 4 | Claude Opus 4.5 Pro | Anthropic | 95.2 | $20.98 | 200,000 |
| 5 | Claude Sonnet 4.5 Reasoning | Anthropic | 95.2 | $19.48 | 400,000 |
| 6 | Grok 4 Reasoning | xAI | 95.2 | $19.67 | 2,000,000 |
| 7 | GPT-5.2 Vision | OpenAI | 94.6 | $19.03 | 1,000,000 |
| 8 | Gemini 3 Pro Turbo | Google DeepMind | 94.6 | $19.62 | 400,000 |
| 9 | GPT-5.1 Reasoning | OpenAI | 94.4 | $20.27 | 400,000 |
| 10 | o4 Pro | OpenAI | 94.4 | $19.37 | 1,000,000 |
| 11 | Grok 4 Max | xAI | 94.1 | $19.70 | 2,000,000 |
| 12 | Claude Sonnet 4.5 Turbo | Anthropic | 93.9 | $18.37 | 1,000,000 |
| 13 | Grok 4 | xAI | 93.9 | $17.90 | 200,000 |
| 14 | Claude Opus 4.5 Reasoning | Anthropic | 93.8 | $20.91 | 2,000,000 |
| 15 | GPT-5.1 Pro | OpenAI | 93.8 | $19.77 | 200,000 |
| 16 | GPT-4.1 Pro | OpenAI | 93.7 | $17.56 | 400,000 |
| 17 | Gemini 3 Pro Pro | Google DeepMind | 93.6 | $19.80 | 200,000 |
| 18 | Gemini 3 Pro Reasoning | Google DeepMind | 93.5 | $19.69 | 200,000 |
| 19 | o3 Max | OpenAI | 93.4 | $17.82 | 2,000,000 |
| 20 | Claude Sonnet 4.5 Max | Anthropic | 93.1 | $19.90 | 1,000,000 |
Methodology
This story uses the AI & LLM Index dataset: 600 models, 40 providers and 12 benchmarks. Prices are listed per million tokens. Scores are modelled estimates on a 0–100 scale and are not official benchmark results. Figures are rounded for readability.
Related
FAQ
What does the HellaSwag: scores, leaders and methodology data show?
HellaSwag leaderboard data from the index. Commonsense sentence completion with adversarial wrong answers. The top score is 96.1 and the average is 81.4.
Are the numbers official?
No. Prices and scores are compiled for this demo index. Benchmark values are modelled estimates, not official results. Verify figures with the provider or benchmark owner before making decisions.
How should I use this ranking?
Use it to form a shortlist. Check the specific model pages, compare price and context, and test the shortlist on your own examples before committing.