When2Speak: leaderboard
Large Language Models (LLMs) excel at generating contextually appropriate responses but remain poorly calibrated for multi-party conversations, where deciding when to speak is as critical as what to say. In such settings, naively responding at every turn leads to excessive interruptions and degraded conversational coherence. We introduce When2Speak, a grounded synthetic dataset and four-stage gene
Metric: 0-shot Macro F1 (self-reported). Source: benchmarklist.com. Saturation forecast: Estimated already saturated. 6 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-4o Mini | 56.9 | #599 |
| 2 | GPT-4.1 Mini | 49.9 | #355 |
| 3 | Llama 3.1 8B | 33.7 | #1151 |
| 4 | Llama 3.3 70B | 32.5 | #579 |
| 5 | Qwen 3 4B | 20.6 | #829 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=when2speak · How It Works · Data refreshed daily, snapshot 2026-10-20.