AA Speech Agent Arena: leaderboard
Artificial Analysis Speech Agent Arena: blind pairwise preference votes between live voice agents (speech-to-speech models and cascaded agent platforms) in conversations, overall slice, fitted to Elo ratings. Elo, higher is better.
Metric: Elo (blind pairwise preference). Source: artificialanalysis.ai. Saturation forecast: Rough model projection: around 2026. 30 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Cartesia Managed Agents | 1111.49 | |
| 2 | Gemini 3.1 Flash Live Minimal | 1089.15 | |
| 3 | Gemini 3.8 Live | 1073.81 | |
| 4 | Gemini 3.1 Flash Live High | 1061.87 | |
| 5 | Grok Voice 3.0 | 1058.6 | |
| 6 | GPT-Live-1 (Sol, low) | 1036.69 | |
| 7 | GPT-Live-1 (Astra, medium) | 1026.89 | |
| 8 | Phonic 1 (High) | 1011.84 | |
| 9 | GPT-Realtime-1.5 | 1000 | |
| 10 | Grok Voice 2.0 (High) | 990.69 |
Interactive version: theaggregate.ai/benchmark?slug=aa-speech-agent-arena · How It Works · Data refreshed daily, snapshot 2026-11-05.