pfgen-bench - Completion Mode - Truthfulness: leaderboard
Metric: Truthfulness Score. Source: github.com. 314 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | DeepSeek V3 | 0.97 |
| 2 | Qwen 3 235B A22B (Non-reasoning) | 0.95 |
| 3 | plamo-2-8B | 0.94 |
| 4 | sarashina2-70B | 0.92 |
| 5 | c4ai-command-r-plus | 0.91 |
| 6 | Nemotron-4 340B Instruct | 0.91 |
| 7 | sarashina2.2-3B | 0.91 |
| 8 | Mistral Large 2 (Nov) Instruct (2411) | 0.9 |
| 9 | sarashina2-13B | 0.9 |
| 10 | sarashina2.2-3B-instruct-v0.1 | 0.89 |
| 11 | Llama 3.1 405B | 0.89 |
| 12 | llm-jp-3-13B | 0.88 |
| 13 | Qwen 3 32B (Non-reasoning) | 0.88 |
| 14 | sarashina2-7B | 0.87 |
| 15 | Phi-4 | 0.86 |
Interactive version: theaggregate.ai/benchmark?slug=pfgen-bench-completion-mode-truthfulness · How It Works · Data refreshed daily, snapshot 2026-09-19.