vCX-Hard - Tool Calling: leaderboard
vCX-Hard tool-call accuracy on real contact-centre calls, production configuration with reasoning off. Higher is better. Source: retellai.com/blog/vcx-hard-benchmark.
Metric: Tool-Call Accuracy (%). Source: www.retellai.com. Saturation forecast: Around February 2027. 27 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3.1 Pro (Preview) | 85.7 | #64 |
| 2 | Claude Sonnet 4.6 | 80.1 | #94 |
| 3 | GPT-5.2 | 79.6 | #115 |
| 4 | GPT-5.1 | 78.2 | #145 |
| 5 | Claude Sonnet 4.5 | 77.5 | #148 |
| 6 | GLM-5.1 | 77.3 | #109 |
| 7 | GPT-5.5 | 75.6 | #33 |
| 8 | GPT-5.4 | 75.2 | #83 |
| 9 | O4 Mini | 73.9 | #185 |
| 10 | Gemini 2.5 Pro | 72.9 | #159 |
| 11 | Gemini 3 Flash | 71.2 | #103 |
| 12 | Claude Haiku 4.5 | 70.7 | #287 |
| 13 | GPT-5 Mini | 70.6 | #193 |
| 14 | Gemini 3.5 Flash | 69.2 | #66 |
| 15 | Gemini 3.1 Flash Lite | 69 | #200 |
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-tool-calling · How It Works · Data refreshed daily, snapshot 2026-11-05.