vCX-Hard - Non-Hallucination (Reasoning): leaderboard
vCX-Hard non-hallucination rate in each model's best configuration with reasoning enabled. Higher is better. Source: retellai.com/blog/vcx-hard-benchmark.
Metric: Non-Hallucination Rate (%). Source: www.retellai.com. Saturation forecast: Around February 2027. 27 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5.5 | 88 | #33 |
| 2 | Gemini 3.1 Pro (Preview) | 83.6 | #64 |
| 3 | GPT-5.4 | 83.3 | #83 |
| 4 | GLM-5.1 | 82.7 | #109 |
| 5 | Claude Sonnet 4.6 | 81.6 | #94 |
| 6 | GPT-5.2 | 81.6 | #115 |
| 7 | GPT-5.1 | 81 | #145 |
| 8 | Gemini 3.5 Flash | 80.7 | #66 |
| 9 | Claude Sonnet 4.5 | 80.6 | #148 |
| 10 | Gemini 3 Flash | 80.1 | #103 |
| 11 | GPT-5 | 78.8 | #105 |
| 12 | Claude Haiku 4.5 | 78.7 | #287 |
| 13 | Gemini 3.1 Flash Lite | 77.7 | #200 |
| 14 | Gemini 2.5 Pro | 77.2 | #159 |
| 15 | Grok 4.3 | 76.3 | #204 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=vcx-hard-non-hallucination-reasoning · How It Works · Data refreshed daily, snapshot 2026-11-05.