Speech Agent Arena BestenlisteArtificial Analysis
Vergleiche Speech-to-Speech-Modelle nach Blindpräferenz in Live-Sprachgesprächen zu Themen wie dem Buchen eines Zahnarzttermins und dem Bestellen von Essen. Weitere Details siehe Überblick und Methodik, oder besuche Modelle und Anbieter vergleichen, um Speech-to-Speech-Modelle in weiteren Benchmarks zu vergleichen.
Range | Creator | Model | Elo | Elo 95% CI | Samples | Task Success Rate | TSR 95% CI | |
|---|---|---|---|---|---|---|---|---|
| 1 | 1-5 |  Google | Gemini 3.1 Flash Live Minimal | 1155 | -57/57 | 151 | — | — |
| 2 | 1-5 |  SpaceXAI | Grok Voice Think Fast 1.0 | 1152 | -56/56 | 158 | — | — |
| 3 | 1-7 |  OpenAI | GPT Realtime (Aug '25) | 1128 | -55/55 | 167 | — | — |
| 4 | 1-7 | Google | Gemini 3.1 Flash Live High | 1119 | -55/55 | 151 | — | — |
| 5 | 2-5 | SpaceXAI | Grok Voice Think Fast 2.0 High | 1116 | -38/38 | 685 | — | — |
| 6 | 3-9 |  Amazon Bedrock | Nova 2.0 Sonic (Mar 2026) | 1078 | -54/54 | 185 | — | — |
| 7 | 4-10 |  Cartesia | Cartesia Line (Default Cascaded System) | 1074 | -54/54 | 165 | — | — |
| 8 | 6-16 | OpenAI | GPT-Realtime-2 (Minimal) | 1043 | -54/54 | 160 | — | — |
| 9 | 6-18 |  Deepgram | Deepgram Voice Agent (Default Cascaded System) | 1035 | -56/56 | 156 | — | — |
| 10 | 7-18 |  ElevenLabs | ElevenLabs Agents (Default Cascaded System) | 1023 | -54/54 | 170 | — | — |
| 11 | 7-18 |  Alibaba Cloud | Qwen3.5 Omni Plus Realtime | 1019 | -57/57 | 131 | — | — |
| 12 | 8-18 | OpenAI | GPT-Realtime-2 (High) | 1013 | -54/54 | 166 | — | — |
| 13 | 8-18 | OpenAI | GPT-Realtime-2.1 High | 1008 | -55/55 | 161 | — | — |
| 14 | 8-18 | Amazon | Nova Sonic | 1003 | -54/54 | 152 | — | — |
| 15 | 15 | OpenAI | GPT-Realtime-1.5 | 1000 | 0/0 | 164 | — | — |
| 16 | 8-18 |  Inworld | Inworld Realtime (Default Cascaded System) | 998 | -52/52 | 176 | — | — |
| 17 | 9-18 | Alibaba Cloud | Qwen3.5 Omni Flash Realtime | 989 | -54/54 | 165 | — | — |
| 18 | 9-18 | OpenAI | GPT-Realtime-2.1 Minimal | 983 | -55/55 | 157 | — | — |
| 19 | 19-22 | Alibaba Cloud | Qwen Audio 3.0 Realtime Flash | 907 | -56/56 | 164 | — | — |
| 20 | 19-22 | OpenAI | GPT-Realtime-2.1 Mini Minimal | 898 | -56/56 | 157 | — | — |
| 21 | 19-22 | Alibaba Cloud | Qwen Audio 3.0 Realtime Plus | 870 | -60/60 | 168 | — | — |
| 22 | 19-22 | OpenAI | GPT Realtime Mini (Oct '25) | 858 | -60/60 | 165 | — | — |