OpenAI GPT-Live-1 API Launch - Full Duplex Bench v1.5 Interactivity: leaderboard
OpenAI's September 2026 GPT-Live-1 API launch evaluation of full-duplex spoken interaction: Full Duplex Bench v 1.5 tests reactions to background speech, side conversations, listener backchannels and interruptions. Interactivity success in percent; higher is better. The release chart does not specify GPT-Live-1's backend or the Realtime comparators' effort, so their source configurations remain distinct.
Metric: Interactivity success (%). Source: openai.com. Saturation forecast: Rough model projection: around 2026. 3 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-Live-1 (API launch, backend unspecified) | 80.1 | |
| 2 | GPT Realtime 2 (API launch, effort unspecified) | 47.8 | |
| 3 | GPT Realtime 2.1 (API launch, effort unspecified) | 45.43 |
Interactive version: theaggregate.ai/benchmark?slug=openai-gpt-live-1-api-launch-full-duplex-bench-v1-5-interactivity · How It Works · Data refreshed daily, snapshot 2026-11-05.