GPT-Realtime-2.1

OpenAI's speech-to-speech models for the Realtime API, with configurable reasoning effort and tool use for voice agents.

Where it ranks

B
Voice agents
Models · Cloud · 941 on Artificial Analysis Speech Agent Arena (models)

Vendor: OpenAI
Family: GPT-Realtime
Released: 6 Jul 2026
Weights: Closed: API and apps only
License: proprietary
API price: $32 in / $64 out per 1M tokens (Audio tokens, USD per 1M (text tokens $4 in / $24 out).); $0.19 per minute of audio (Artificial Analysis 'Cost per Hour of Input Audio' for GPT-Realtime-2.1 Minimal, OpenAI ($11.307/hour: cost of its Big Bench Audio run normalised to input-audio length) divided by 60.)
Last verified: 18 Oct 2026

Price from developers.openai.com, 18 Oct 2026.

Benchmark	Result	Entry on the leaderboard	As of	Source
Artificial Analysis Big Bench Audio (speech reasoning)	95.8%	`GPT-Realtime-2.1 High, OpenAI` Big Bench Audio speech reasoning (audio in, audio out); embedded value 0.958 as percent; no last-updated date shown.	18 Oct 2026	artificialanalysis.ai
Artificial Analysis Speech Agent Arena (models)	941	`GPT-Realtime-2.1 Minimal` Speech Agent Arena Elo (overall slice; GPT-Realtime-1.5 anchored at 1000), rank 14 of 27, 828 appearances, 95% CI 915.24-966.11; embedded data on the speech-to-speech page.	18 Oct 2026	artificialanalysis.ai
Artificial Analysis Speech to Speech Index	75	`GPT-Realtime-2.1 High` AA-Speech to Speech Index (weighted: Big Bench Audio, tau-Voice agentic, arena preference, task success); raw embedded value, page shows it rounded; no last-updated date shown.	18 Oct 2026	artificialanalysis.ai
Cekura Bench: voice agent workflow benchmark v1 (speech-to-speech models)	64.6%	`GPT Realtime` Cekura reference agent on gpt-realtime-2.1; frozen v1 release.	12 Oct 2026	benchmarks.cekura.ai