VAmoS Bench: leaderboard
Veris AI voice-agent simulation benchmark: task completion rate over simulated phone-call flows (simple, complex, adversarial). Rows are voice stacks; scores are kept per underlying LLM, best stack per model.
Metric: Task Completion (%). Source: veris.ai. Saturation forecast: Around February 2027. 10 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-4.1 Mini | 71 | #363 |
| 2 | DeepSeek V4 Flash | 65.3 | #155 |
| 3 | Mistral Small | 59.7 | #731 |
| 4 | Hermes 4 70B | 58.7 | #556 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=vamos-bench · How It Works · Data refreshed daily, snapshot 2026-11-05.