legal benchmark
TruthfulQA Leaderboard
TruthfulQA is a benchmark measuring whether language models generate truthful answers to questions.
Updated Sep 5, 2026
Models18
Model coverage18
MetricScore
EvidenceB
TruthfulQA Ranking
Higher score ranks better on this benchmark.
Sort by
Rank | Model | Score | Percentile | Participants | Evidence | Evaluated |
|---|
| Rank01 | ModelMI | Score88.00% | Percentile100.00% | Participants18 | EvidenceC | Evaluated |
| Rank02 | ModelMI | Score77.50% | Percentile94.12% | Participants18 | EvidenceC | Evaluated |
| Rank03 | ModelIB | Score66.86% | Percentile88.24% | Participants18 | EvidenceC | Evaluated |
| Rank04 | ModelMI | Score66.40% | Percentile82.35% | Participants18 | EvidenceC | Evaluated |
| Rank05 | ModelMI | Score64.00% | Percentile76.47% | Participants18 | EvidenceC | Evaluated |
| Rank06 | ModelNR | Score63.29% | Percentile70.59% | Participants18 | EvidenceC | Evaluated |
| Rank07 | ModelNV | Score58.63% | Percentile64.71% | Participants18 | EvidenceC | Evaluated |
| Rank08 | ModelAC | Score58.40% | Percentile58.82% | Participants18 | EvidenceC | Evaluated |
| Rank09 | ModelAL | Score58.30% | Percentile52.94% | Participants18 | EvidenceC | Evaluated |
| Rank10 | ModelIB | Score58.10% | Percentile47.06% | Participants18 | EvidenceC | Evaluated |
| Rank11 | ModelAC | Score57.80% | Percentile41.18% | Participants18 | EvidenceC | Evaluated |
| Rank12 | ModelCO | Score56.30% | Percentile35.29% | Participants18 | EvidenceC | Evaluated |
| Rank13 | ModelAC | Score54.80% | Percentile29.41% | Participants18 | EvidenceC | Evaluated |
| Rank14 | ModelAC | Score54.20% | Percentile23.53% | Participants18 | EvidenceC | Evaluated |
| Rank15 | ModelAL | Score54.10% | Percentile17.65% | Participants18 | EvidenceC | Evaluated |
| Rank16 | ModelIB | Score52.15% | Percentile11.76% | Participants18 | EvidenceC | Evaluated |
| Rank17 | ModelAC | Score50.60% | Percentile5.88% | Participants18 | EvidenceC | Evaluated |
| Rank18 | ModelMA | Score50.30% | Percentile0.00% | Participants18 | EvidenceC | Evaluated |
TruthfulQA Highlights
The leading models and scores on this benchmark.
TruthfulQA Score Distribution
A closer view of the leading scores on this benchmark.
The Top AI Models for TruthfulQA
The first five results on this benchmark, with official price and output speed added where the model identity can be matched.
What is TruthfulQA?
What TruthfulQA measures and how its scores work.
TruthfulQA comprises 817 questions across 38 categories, including health, law, finance and politics.
It measures models' ability to avoid generating false answers associated with human false beliefs, misconceptions and texts.
Scores are shown in ratio. This benchmark is not independently verified and has an evidence level of B.
- Family
- TruthfulQA
- Modality
- text
- Primary category
- legal
- Score direction
- higher
- LLMBoard eligible
- No
- Evaluation key
- overall
LLM Stats. Benchmark scores retain their original unit. Overall score eligibility is shown separately.
FAQ
Common questions about TruthfulQA.
Which model scores highest on TruthfulQA?
MAI-Thinking-1 is currently ranked first with 88.00%.
What are the top three models on TruthfulQA?
The current leaders are MAI-Thinking-1 (88.00%), Phi-3.5-MoE-instruct (77.50%), and Granite 3.3 8B Instruct (66.86%).
Which TruthfulQA model has the lowest official input price?
Phi 4 Mini has the lowest matched official input price at $0.08 input / $0.30 output per 1M tokens.
Which models are fastest among TruthfulQA results?
The fastest matched records are Jamba 1.5 Mini (100.00 tok/s via Google), Command R+ (59.00 tok/s via Cohere), and Qwen2.5-Coder 32B Instruct (44.00 tok/s via DeepInfra).
Does the highest score result prove overall model quality?
No. This benchmark measures one defined capability or task. The overall LLMBoard score uses a separate aggregation across eligible benchmark evidence.