OpenAIAudio matchupReal-time voice dialog
GPT-Realtime 2.1vs.GPT-Live 1
Which model is better?
GPT-Realtime 2.1 and GPT-Live 1 each lead in different tests.
Pick GPT-Realtime 2.1 for
Pick GPT-Live 1 for
Shared tests
Independent evaluators and leaderboards only, the newest shared snapshot per evaluator. With 95 % ranges a lead counts when they do not overlap; for hit rates, when it holds in a 95 % test. Without any uncertainty figure at most “slightly ahead”.
Speko
depends on the subtestGPT-Live 1 is clearly ahead in 0 of 7 subtests and slightly in 3, GPT-Realtime 2.1 in 0 clearly and slightly in 2. 2 are level.
Speko S2S task-content latency p50
lower is better, as of October 28, 2026
Speko S2S first-sound latency p50
lower is better, as of October 28, 2026
Speko S2S confirmation-safe action success
as of October 28, 2026
Show all 7 subtests
| Subtest | GPT-Realtime 2.1 | GPT-Live 1 | Verdict |
|---|---|---|---|
| Speko S2S task-content latency p50 (lower is better) | 1,145 ms | 1,892 ms | GPT-Realtime 2.1 slightly ahead |
| Speko S2S first-sound latency p50 (lower is better) | 1,022 ms | 1,482 ms | GPT-Realtime 2.1 slightly ahead |
| Speko S2S confirmation-safe action success | 78.0 | 90.0 | GPT-Live 1 slightly ahead |
| Speko S2S overall call quality | 84.0 | 88.0 | GPT-Live 1 slightly ahead |
| Speko S2S task capability mean | 86.0 | 90.0 | GPT-Live 1 slightly ahead |
| Speko S2S scripted call completion | 100.0 % | 98.0 % | level |
| Speko S2S booking before caller confirmation (lower is better) | 0.0 % | 0.0 % | level |
Source · latest October 28, 2026
Cekura
GPT-Realtime 2.1 slightly aheadGPT-Realtime 2.1 is clearly ahead in 0 of 4 subtests and slightly in 1, GPT-Live 1 in none. 3 are level.
Cekura TurnBench end-of-turn score, Loud noise
36 tasks, as of October 23, 2026
Cekura TurnBench end-of-turn score, Light noise
36 tasks, as of October 23, 2026
Cekura TurnBench end-of-turn score, Echoey rooms
36 tasks, as of October 23, 2026
Show all 4 subtests
| Subtest | GPT-Realtime 2.1 | GPT-Live 1 | Verdict |
|---|---|---|---|
| Cekura TurnBench end-of-turn score, Loud noise | 84.0 % | 81.0 % | GPT-Realtime 2.1 slightly ahead |
| Cekura TurnBench end-of-turn score, Light noise | 83.0 % | 85.0 % | level |
| Cekura TurnBench end-of-turn score, Echoey rooms | 88.0 % | 86.0 % | level |
| Cekura TurnBench end-of-turn score, six-condition average | 85.0 % | 84.0 % | level |
Source · latest October 23, 2026
Artificial Analysis
GPT-Live 1 slightly aheadArtificial Analysis Speech Agent Arena preference Elo
as of October 28, 2026
Source · latest October 28, 2026
All data side by side
Prices are verbatim from the provider, because audio models bill per minute, character or token.
| GPT-Realtime 2.1 | GPT-Live 1 | |
|---|---|---|
| Task | Real-time voice dialog | Real-time voice dialog |
| Developer | OpenAI | OpenAI |
| Status | Active | Active |
| Released | n/a | September 10, 2026 |
| Weights | proprietary | n/a |
| Runs locally | no | no |
| Price | $4 input, $0.40 cached input, $24 output per 1M tokens | $0.05 per voice-session minute, billed per second. Backend model and tools cost extra. |
| Popularity (out of 100) | 41 | 0 |