Best AI for Verifiability (November 2026)
Data refreshed 1 Nov 2026
d1-ultra leads ClaimBench with 100 (accuracy %). Claude Sonnet 4 leads FAITH with 95.6 (overall accuracy %). Gemini 2.5 Flash leads K-FinHallu with 0.860 (binary detection F1). GPT-4o leads FinLFQA with 56.5 (evidence attribution F1).
- Benchmarks in this use case
- 5
- With dated scores
- 4
- Models with scores
- 33
- Latest source date
- 23 Oct 2026
Verifiability picks
The leader on each benchmark, never combinedBest on ClaimBench
d1-ultraMaker not stated
100 accuracy %
Official board, 23 Oct 2026
Best on FAITH
Claude Sonnet 4Anthropic
95.6 overall accuracy %
Reported by the authors, 24 Oct 2025
Best on K-FinHallu
Gemini 2.5 FlashGoogle
0.860 binary detection F1
Reported by the authors, 28 May 2026
Best on FinLFQA
GPT-4oOpenAI
56.5 evidence attribution F1
Reported by the authors, 7 Oct 2025
Verifiability leaderboard
ClaimBench v1 (updated 2026-10-23)
accuracy %, higher is better. Top 5 of 10 models.
Updated 23 Oct 2026
- 1d1-ultra100
- 2Decider V1.299.7
- 3Drex Max99.2
- 4Mercury Decide99.1
- 5Decider V1.1 27B98.8
| 1 | d1-ultra | Not stated | 100 | decision models board | Vals.ai leaderboard Official board, 23 Oct 2026 |
|---|---|---|---|---|---|
| 2 | Decider V1.2 | Not stated | 99.7 | decision models board | Vals.ai leaderboard Official board, 23 Oct 2026 |
| 3 | Drex Max | Not stated | 99.2 | decision models board | Vals.ai leaderboard Official board, 23 Oct 2026 |
| 4 | Mercury Decide | Not stated | 99.1 | decision models board | Vals.ai leaderboard Official board, 23 Oct 2026 |
| 5 | Decider V1.1 27B | Not stated | 98.8 | decision models board | Vals.ai leaderboard Official board, 23 Oct 2026 |
| 6 | d1 | Not stated | 98.7 | decision models board | Vals.ai leaderboard Official board, 23 Oct 2026 |
| 7 | Microsoft-Decision-1 | Microsoft | 98.2 | decision models board | Vals.ai leaderboard Official board, 23 Oct 2026 |
| 8 | Jev 1.14 | Not stated | 98.1 | decision models board | Vals.ai leaderboard Official board, 23 Oct 2026 |
| 9 | Drex v1.5 | Not stated | 98.0 | decision models board | Vals.ai leaderboard Official board, 23 Oct 2026 |
| 10 | Jev 1.13.0 | Not stated | 97.7 | decision models board | Vals.ai leaderboard Official board, 23 Oct 2026 |
FAITH arXiv v2 (2025-10-24), ACM ICAIF'25
overall accuracy %, higher is better. Top 5 of 10 models.
Updated 24 Oct 2025
- 1Claude Sonnet 495.6
- 2Gemini 2.5 Pro91.9
- 3GPT-4.189.2
- 4Gemini 2.5 Flash88.7
- 5GPT-4.1-mini88.2
| 1 | Claude Sonnet 4 | Anthropic | 95.6 | Main split | Paper Table 3 Reported by the authors, 24 Oct 2025 |
|---|---|---|---|---|---|
| 2 | Gemini 2.5 Pro | 91.9 | Main split | Paper Table 3 Reported by the authors, 24 Oct 2025 | |
| 3 | GPT-4.1 | OpenAI | 89.2 | Main split | Paper Table 3 Reported by the authors, 24 Oct 2025 |
| 4 | Gemini 2.5 Flash | 88.7 | Main split | Paper Table 3 Reported by the authors, 24 Oct 2025 | |
| 5 | GPT-4.1-mini | OpenAI | 88.2 | Main split | Paper Table 3 Reported by the authors, 24 Oct 2025 |
| 6 | Claude-haiku-3 | Anthropic | 81.3 | Main split | Paper Table 3 Reported by the authors, 24 Oct 2025 |
| 7 | Qwen-3-32B | Qwen | 73.9 | Main split | Paper Table 3 Reported by the authors, 24 Oct 2025 |
| 8 | GPT-4.1-nano | OpenAI | 70 | Main split | Paper Table 3 Reported by the authors, 24 Oct 2025 |
| 9 | Gemini-2.5-flash-lite | 50.2 | Main split | Paper Table 3 Reported by the authors, 24 Oct 2025 | |
| 10 | Llama-3.1-8B | Meta | 47.5 | Main split | Paper Table 3 Reported by the authors, 24 Oct 2025 |
K-FinHallu arXiv v1 (2026-05-28)
binary detection F1, higher is better. Top 5 of 10 models.
Updated 28 May 2026
- 1Gemini 2.5 Flash0.860
- 2Gemini 2.5 Pro0.849
- 3gpt-50.823
- 4Llama-3.3-70B0.756
- 5Llama-3.1-70B0.735
| 1 | Gemini 2.5 Flash | 0.860 | test split | Paper Table 3 Reported by the authors, 28 May 2026 | |
|---|---|---|---|---|---|
| 2 | Gemini 2.5 Pro | 0.849 | test split | Paper Table 3 Reported by the authors, 28 May 2026 | |
| 3 | gpt-5 | OpenAI | 0.823 | test split | Paper Table 3 Reported by the authors, 28 May 2026 |
| 4 | Llama-3.3-70B | Meta | 0.756 | test split | Paper Table 3 Reported by the authors, 28 May 2026 |
| 5 | Llama-3.1-70B | Meta | 0.735 | test split | Paper Table 3 Reported by the authors, 28 May 2026 |
| 6 | GPT-4o | OpenAI | 0.733 | test split | Paper Table 3 Reported by the authors, 28 May 2026 |
| 7 | Qwen3-32B | Qwen | 0.732 | test split | Paper Table 3 Reported by the authors, 28 May 2026 |
| 8 | EXAONE-4.0-32B | LG | 0.728 | test split | Paper Table 3 Reported by the authors, 28 May 2026 |
| 9 | Qwen3-14B | Qwen | 0.707 | test split | Paper Table 3 Reported by the authors, 28 May 2026 |
| 10 | Llama-3.1-8B | Meta | 0.674 | test split | Paper Table 3 Reported by the authors, 28 May 2026 |
FinLFQA arXiv v1 (2025-10-07), Findings of EMNLP 2025
evidence attribution F1, higher is better. Top 5 of 8 models.
Updated 7 Oct 2025
- 1GPT-4o56.5
- 2Mistral-Small-24B55.3
- 3Llama-3.3-70B52.9
- 4Qwen2.5-72B46.6
- 5phi-446.1
| 1 | GPT-4o | OpenAI | 56.5 | post-hoc attribution, test set | Paper Table 6 Reported by the authors, 7 Oct 2025 |
|---|---|---|---|---|---|
| 2 | Mistral-Small-24B | Mistral AI | 55.3 | post-hoc attribution, test set | Paper Table 6 Reported by the authors, 7 Oct 2025 |
| 3 | Llama-3.3-70B | Meta | 52.9 | post-hoc attribution, test set | Paper Table 6 Reported by the authors, 7 Oct 2025 |
| 4 | Qwen2.5-72B | Qwen | 46.6 | post-hoc attribution, test set | Paper Table 6 Reported by the authors, 7 Oct 2025 |
| 5 | phi-4 | Microsoft | 46.1 | post-hoc attribution, test set | Paper Table 6 Reported by the authors, 7 Oct 2025 |
| 6 | Mistral-8x22B | Mistral AI | 42.9 | post-hoc attribution, test set | Paper Table 6 Reported by the authors, 7 Oct 2025 |
| 7 | Llama-3.2-3B | Meta | 11.8 | post-hoc attribution, test set | Paper Table 6 Reported by the authors, 7 Oct 2025 |
| 8 | Llama-3.2-1B | Meta | 1.5 | post-hoc attribution, test set | Paper Table 6 Reported by the authors, 7 Oct 2025 |