TrustGen - TrustLLM External: leaderboard
TrustGen - TrustLLM External. TrustGen paper 2502.14296v2 provides native accuracy measurements and exact model versions in Table 5. Internal question answering uses parametric knowledge; fact checking classifies claims against supplied evidence rather than using a live retrieval system. Dynamic QA comes from Wikipedia and dynamic fact checking from fact-checking websites, with varied prompt formats. TrustLLM columns use the paper’s dataset pool and LLM judging against reference answers. HallusionBench and AutoHallusion are the paper’s visual-question evaluations, with their reported subsets kept separate. These source-specific runs are not automatically fused with another publisher’s results. QwQ-32B is held because the table does not distinguish its preview or released checkpoint. Reported metric: Accuracy (%).
Metric: Accuracy (%). Source: arxiv.org. Saturation forecast: Estimated already saturated. 18 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-3.5 Turbo (0125) | 74.25 | #984 |
| 2 | Claude 3 Haiku (20240307) | 70 | #781 |
| 3 | Llama 3.1 8B Instruct | 69.75 | #1044 |
| 4 | Mixtral 8x22B Instruct (v0.1) | 67.25 | #755 |
| 5 | DeepSeek V2.5 | 66.75 | #490 |
| 6 | Qwen 2.5 72B Instruct | 65.25 | #463 |
| 7 | O1 Mini (2024-09-12) | 64.41 | #411 |
| 8 | Yi Lightning | 63.75 | #423 |
| 9 | GPT-4o Mini (2024-07-18) | 63.25 | #562 |
| 10 | O1 Preview (2024-09-12) | 62.12 | #215 |
| 11 | Gemma 2 27B (IT) | 61 | #653 |
| 12 | GLM-4 Plus | 59.25 | #397 |
| 13 | Llama 3.1 70B Instruct | 56.25 | #568 |
| 14 | Claude 3.5 Sonnet (20240620) | 54.25 | #305 |
| 15 | Gemini 1.5 Flash (002) | 54.14 | #589 |
No result here: #5 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=trustgen-trustllm-external · How It Works · Data refreshed daily, snapshot 2026-10-25.