pfgen-bench - Completion Mode - Helpfulness: leaderboard
Metric: Helpfulness Score. Source: github.com. 314 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | DeepSeek V3 | 0.82 |
| 2 | Qwen 3 235B A22B (Non-reasoning) | 0.59 |
| 3 | sarashina2.2-3B-instruct-v0.1 | 0.52 |
| 4 | sarashina2-70B | 0.49 |
| 5 | c4ai-command-r-plus | 0.48 |
| 6 | sarashina2.2-3B | 0.47 |
| 7 | plamo-2-8B | 0.46 |
| 8 | Nemotron-4 340B Instruct | 0.43 |
| 9 | Mistral Large 2 (Nov) Instruct (2411) | 0.42 |
| 10 | sarashina2-13B | 0.39 |
| 11 | Qwen 3 32B (Non-reasoning) | 0.38 |
| 12 | Llama 3.1 405B | 0.37 |
| 13 | llm-jp-3-13B | 0.35 |
| 14 | Qwen 2.5 72B | 0.34 |
| 15 | sarashina2-7B | 0.32 |
Interactive version: theaggregate.ai/benchmark?slug=pfgen-bench-completion-mode-helpfulness · How It Works · Data refreshed daily, snapshot 2026-09-19.