DROP — DROP F1 metric, no shot-count qualifier; kept separate from 3-shot F1 (diff source/range)
| # | Model | Vendor | Best score | Runs | Last seen |
|---|---|---|---|---|---|
| 1 | Llama 3.1 Instruct 405B | Meta | 92.5 | 2 | 2026-06-06 |
| 2 | DeepSeek-V3 | DeepSeek | 91 | 2 | 2026-06-06 |
| 3 | Gemini 2.0 Flash Exp | 89.3 | 1 | 2026-06-06 | |
| 4 | GPT-4o | OpenAI | 89.2 | 1 | 2026-06-06 |
| 5 | Gemini 1.5 Pro | 89.2 | 1 | 2026-06-06 | |
| 6 | Claude 3.5 Sonnet | Anthropic | 88.8 | 1 | 2026-06-06 |
| 7 | DeepSeek-V4-Pro-Base | DeepSeek | 88.7 | 4 | 2026-06-27 |
| 8 | DeepSeek-V4-Flash-Base | DeepSeek | 88.6 | 4 | 2026-06-27 |
| 9 | DeepSeek-V3.2-Base | DeepSeek | 88.2 | 4 | 2026-06-27 |
| 10 | MiniMax Text 01 | MiniMax | 87.8 | 1 | 2026-06-06 |
| 11 | Qwen2.5 Instruct 72B | Alibaba | 85 | 2 | 2026-06-06 |
| 12 | DeepSeek-V2 | DeepSeek | 80.4 | 1 | 2026-05-01 |
Best tracked score per model (default configuration; source-attributed and verification-tiered). Open a model for its full benchmark surface, provenance and pricing.