Liquid LFM2.5-2.6B Card - AA-Omniscience-Public Non-Hallucination Rate: leaderboard
Non-hallucination rate on the public question set of Artificial Analysis' Omniscience evaluation: among questions a model does not answer correctly, the share where it avoids a wrong answer, for example by abstaining. Run by Liquid AI with vLLM for the LFM2.5-2.6B model card; percent, higher is better.
Metric: Non-Hallucination Rate (%). Source: huggingface.co. Saturation forecast: Not forecast. 5 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | LFM2.5-2.6B | 59.04 | #836 |
| 2 | gemma-4-E4B-it | 37.42 | #577 |
| 3 | Gemma 4 E2B (IT) | 13.67 | #779 |
| 4 | Qwen 3.5 4B | 12.66 | #514 |
| 5 | Qwen 3.5 9B | 8.84 | #391 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=liquid-lfm2-5-2-6b-card-aa-omniscience-public-non-hallucination-rate · How It Works · Data refreshed daily, snapshot 2026-11-05.