Vectara HHEM leaderboard — Vectara hallucination-eval leaderboard, answer-rate metric column
| # | Model | Vendor | Best score | Runs | Last seen |
|---|---|---|---|---|---|
| 1 | GPT-5.1 | OpenAI | 100 | 4 | 2026-08-10 |
| 2 | Gemma 4 31B | 100 | 1 | 2026-05-02 | |
| 3 | GPT-5.2 | OpenAI | 100 | 4 | 2026-06-22 |
| 4 | GPT-5.4 nano | OpenAI | 100 | 1 | 2026-05-02 |
| 5 | o3-pro | OpenAI | 100 | 1 | 2026-05-02 |
| 6 | Llama 4 Maverick 17B 128E Instruct FP8 | Meta | 100 | 1 | 2026-05-02 |
| 7 | Granite 3.3 8B Instruct | IBM | 100 | 1 | 2026-05-02 |
| 8 | GPT-5.4 mini | OpenAI | 100 | 1 | 2026-05-02 |
| 9 | GPT-5 nano | OpenAI | 100 | 2 | 2026-06-18 |
| 10 | Nova Micro (Non-Reasoning) | Amazon | 100 | 1 | 2026-05-02 |
| 11 | GPT-5.4 Pro | OpenAI | 100 | 1 | 2026-05-02 |
| 12 | GPT-5.5 | OpenAI | 100 | 1 | 2026-05-02 |
| 13 | Granite 4.0 H Small | IBM | 100 | 2 | 2026-06-19 |
| 14 | Mercury 2 | Inception | 100 | 1 | 2026-05-02 |
| 15 | GPT-4.1 | OpenAI | 99.9 | 1 | 2026-05-02 |
| 16 | Qwen3 32B | Alibaba | 99.9 | 1 | 2026-05-02 |
| 17 | Claude Sonnet 4.6 | Anthropic | 99.9 | 1 | 2026-05-02 |
| 18 | Nova Lite (Non-Reasoning) | Amazon | 99.9 | 1 | 2026-05-02 |
| 19 | GPT-5.4 | OpenAI | 99.9 | 1 | 2026-05-02 |
| 20 | GPT-5 | OpenAI | 99.9 | 4 | 2026-08-10 |
| 21 | Mistral Large 2.1 | Mistral | 99.9 | 1 | 2026-05-02 |
| 22 | GPT-5 mini | OpenAI | 99.9 | 1 | 2026-05-02 |
| 23 | Qwen3 8B | Alibaba | 99.9 | 1 | 2026-05-02 |
| 24 | Qwen3 4B | Alibaba | 99.9 | 1 | 2026-05-02 |
| 25 | Ministral 8B 2410 | Mistral | 99.9 | 1 | 2026-05-02 |
| 26 | Qwen3 14B | Alibaba | 99.9 | 1 | 2026-05-02 |
| 27 | Ministral 3B 2410 | Mistral | 99.9 | 1 | 2026-05-02 |
| 28 | GPT OSS 120B | OpenAI | 99.9 | 1 | 2026-05-02 |
| 29 | Qwen3.5 Flash | Alibaba | 99.8 | 1 | 2026-05-02 |
| 30 | c4ai-aya-expanse-32b | Cohere | 99.8 | 1 | 2026-05-02 |
| 31 | Qwen3.5 35B A3B | Alibaba | 99.8 | 2 | 2026-05-04 |
| 32 | Qwen3.5 122B A10B | Alibaba | 99.8 | 1 | 2026-05-02 |
| 33 | GLM-4.7 | Z.ai | 99.8 | 1 | 2026-05-02 |
| 34 | Gemma 4 26B A4B IT | 99.8 | 1 | 2026-05-02 | |
| 35 | Qwen3.5 Plus 02.15 | Alibaba | 99.8 | 1 | 2026-05-02 |
| 36 | Gemini 3 Flash Preview | 99.8 | 1 | 2026-05-02 | |
| 37 | Claude Opus 4.6 | Anthropic | 99.8 | 1 | 2026-05-02 |
| 38 | Qwen3.5 27B | Alibaba | 99.8 | 1 | 2026-05-02 |
| 39 | Grok 4.1 Fast | SpaceXAI | 99.7 | 2 | 2026-05-02 |
| 40 | Mistral Medium 3.1 (Non-Reasoning) | Mistral | 99.7 | 2 | 2026-06-23 |
| 41 | GLM-5 | Z.ai | 99.7 | 1 | 2026-05-02 |
| 42 | Kimi K2.6 | Moonshot | 99.7 | 1 | 2026-05-12 |
| 43 | Ministral 3 14B | Mistral | 99.6 | 1 | 2026-05-02 |
| 44 | Gemini 3.1 Flash Lite Preview | 99.6 | 1 | 2026-05-02 | |
| 45 | Jamba Mini 2 | AI21 | 99.6 | 1 | 2026-05-02 |
| 46 | Nova 2.0 Lite | Amazon | 99.6 | 1 | 2026-05-02 |
| 47 | NVIDIA Nemotron 3 Nano 30B A3B | NVIDIA | 99.6 | 1 | 2026-05-02 |
| 48 | Llama 3.3 70B Instruct | Meta | 99.5 | 1 | 2026-05-02 |
| 49 | Grok 4 Fast | SpaceXAI | 99.5 | 2 | 2026-05-02 |
| 50 | Gemini 2.5 Flash Lite | 99.5 | 1 | 2026-05-02 | |
| 51 | Claude Haiku 4.5 | Anthropic | 99.5 | 1 | 2026-05-02 |
| 52 | finix_s1_32b | antgroup | 99.5 | 1 | 2026-05-02 |
| 53 | Gemini 3.1 Pro | 99.4 | 1 | 2026-05-02 | |
| 54 | MiniMax M2.7 | MiniMax | 99.4 | 1 | 2026-05-02 |
| 55 | Gemini 3 Pro | 99.4 | 1 | 2026-05-02 | |
| 56 | Nova Pro (Non-Reasoning) | Amazon | 99.3 | 1 | 2026-05-02 |
| 57 | o4-mini | OpenAI | 99.2 | 2 | 2026-05-02 |
| 58 | Ministral 3 8B | Mistral | 99.1 | 1 | 2026-05-02 |
| 59 | Gemini 2.5 Pro | 99.1 | 1 | 2026-05-02 | |
| 60 | Jamba Mini 1.7 2025-07 | AI21 | 99.1 | 1 | 2026-05-02 |
| 61 | Gemini 2.5 Flash | 99 | 1 | 2026-05-02 | |
| 62 | Trinity Large Thinking | Arcee AI | 99 | 1 | 2026-05-02 |
| 63 | Llama 4 Scout | Meta | 99 | 1 | 2026-05-02 |
| 64 | Jamba Large 1.7 2025-07 | AI21 | 98.9 | 1 | 2026-05-02 |
| 65 | Gemma 3 27B | 98.8 | 1 | 2026-05-02 | |
| 66 | Mistral Large 3 | Mistral | 98.8 | 1 | 2026-05-02 |
| 67 | Claude Opus 4.5 | Anthropic | 98.7 | 2 | 2026-06-20 |
| 68 | Kimi K2 (Non-Reasoning) | Moonshot | 98.6 | 1 | 2026-06-20 |
| 69 | Kimi K2 Instruct | Moonshot | 98.6 | 1 | 2026-07-06 |
| 70 | Claude Sonnet 4-20250514 (no thinking) | Anthropic | 98.6 | 1 | 2026-06-20 |
| 71 | Kimi K2 (Reasoning) | Moonshot | 98.6 | 1 | 2026-05-02 |
| 72 | Claude Sonnet 4 | Anthropic | 98.6 | 1 | 2026-05-02 |
| 73 | MiniMax M2.1 | MiniMax | 98.5 | 1 | 2026-05-02 |
| 74 | MiniMax M2.5 | MiniMax | 98.2 | 1 | 2026-05-02 |
| 75 | GLM-4.5-Air | Z.ai | 98.1 | 1 | 2026-05-02 |
| 76 | Claude Opus 4.7 | Anthropic | 98 | 1 | 2026-05-02 |
| 77 | Mistral Small 3 (Non-Reasoning) | Mistral | 97.9 | 1 | 2026-05-02 |
| 78 | Command A 03-2025 | Cohere | 97.6 | 1 | 2026-05-02 |
| 79 | DeepSeek-V3 | DeepSeek | 97.5 | 1 | 2026-05-02 |
| 80 | Gemma 3 12B | 97.4 | 1 | 2026-05-02 | |
| 81 | DeepSeek-V4-Pro | DeepSeek | 97.2 | 1 | 2026-06-29 |
| 82 | DeepSeek-R1 | DeepSeek | 97 | 1 | 2026-05-02 |
| 83 | DeepSeek-V3.2-Exp | DeepSeek | 96.6 | 1 | 2026-05-02 |
| 84 | Claude Sonnet 4.5 | Anthropic | 95.6 | 2 | 2026-05-30 |
| 85 | Command R+ (Apr '24) | Cohere | 95 | 1 | 2026-05-02 |
| 86 | Qwen3 235B A22B | Alibaba | 94.9 | 1 | 2026-05-02 |
| 87 | DeepSeek-V3.1 | DeepSeek | 94.5 | 1 | 2026-05-02 |
| 88 | GLM-4.6 | Z.ai | 94.5 | 1 | 2026-05-02 |
| 89 | Qwen3 Next 80B A3B | Alibaba | 94.4 | 1 | 2026-05-02 |
| 90 | GPT-4o | OpenAI | 93.8 | 1 | 2026-05-02 |
| 91 | Grok 3 | SpaceXAI | 93 | 2 | 2026-06-19 |
| 92 | DeepSeek-V3.2 | DeepSeek | 92.6 | 1 | 2026-05-02 |
| 93 | Phi 4 Mini Instruct | Microsoft | 92.5 | 2 | 2026-06-19 |
| 94 | Claude Opus 4.1 | Anthropic | 92.4 | 2 | 2026-06-18 |
| 95 | Kimi K2.5 | Moonshot | 92.2 | 1 | 2026-05-02 |
| 96 | GLM-4.7-Flash | Z.ai | 91.6 | 1 | 2026-05-02 |
| 97 | Claude 4 Opus | Anthropic | 91 | 2 | 2026-06-20 |
| 98 | Phi 4 | Microsoft | 80.7 | 1 | 2026-05-02 |
| 99 | c4ai-aya-expanse-8b | Cohere | 77.5 | 1 | 2026-05-02 |
| 100 | Gemma 3 4B | 67.3 | 1 | 2026-05-02 |
Best tracked score per model (default configuration; source-attributed and verification-tiered). Open a model for its full benchmark surface, provenance and pricing.