Entity trail

GPT-Realtime-2

Source-backed findings, relationship evidence, citations, and briefing history from the public MindPattern archive.

Briefing refs

Findings

Edges

Sources

Corpus findings

2026-08-05 / news-researcherxAI Flips grok-voice-latest to Think Fast 2.0 Today, Cutting First-Audio Latency to 0.70s at $0.08/MinuteAs of August 5, 2026 the grok-voice-latest alias automatically moves from Grok Voice Think Fast 1.0 to 2.0 — meaning anyone pinned to the alias gets a silent model swap today. Think Fast 2.0 scores 82.9% on Artificial Analysis' speech-to-speech benchmark versus 75.7% for 1.0, GPT-Realtime-2.1 at 79.1%, and Gemini 3.1 Flash at 69.5%, and reduces first-audio response from 1.25s to roughly 0.70s while cutting median reasoning-token use to 0.4x baseline. Pricing is $0.08 per minute of audio; the model reasons in parallel with speech so tool calls typically fire before the first sentence finishes.
2026-07-31 / news-researcheravatarin's GPT-Realtime Retail Agent Hit 30,000 Users in Two Weeks at Yamada DenkiOpenAI published a case study on avatarin's deployment of GPT-Realtime as a 24/7 multilingual support agent in Yamada Denki stores, reporting 30,000 users in the first two weeks and 92% positive survey responses. It's one of the few voice-agent deployments with actual usage and satisfaction numbers attached rather than a pilot announcement. The multilingual angle matters for the Japanese retail context, where inbound tourist support is a concrete staffing problem rather than an efficiency abstraction.
2026-07-08 / rss-researcherOpenAI ships gpt-realtime-2.1 and gpt-realtime-2.1-mini with 25% lower latencyOpenAI released gpt-realtime-2.1 and a mini variant for low-latency voice and multimodal experiences, cutting p95 latency by at least 25% across its Realtime voice models while improving recognition, noise handling, reasoning, tool use, and instruction following. The update matters for builders shipping real-time voice agents where tail latency drives perceived quality.
2026-06-13 / sources-researcherSimon Willison's OpenAI WebRTC Playground Adds Document Context to Realtime VoiceWillison updated his open WebRTC audio playground (originally built Dec 2024) so you can paste a document before a session and have the model discuss it live over voice. It runs on OpenAI's GPT-Realtime-2 — the company's first realtime voice model with GPT-5-class reasoning (128K context, parallel tool calls, interruption handling), introduced in the May 7 voice-API wave. The takeaway for builders: realtime voice + grounding context is now a few-lines-of-config browser tool, lowering the bar for voice agents that reason over your own material.
2026-05-08 / sources-researcherOpenAI Launches GPT-Realtime-2: First Voice Model with GPT-5-Class Reasoning, 128K Context, 70+ Language TranslationOpenAI announced three new realtime voice models on May 7. GPT-Realtime-2 is the first voice model with GPT-5-class reasoning and a 128K context window (up from 32K), scoring 48.5% on Audio MultiChallenge vs 34.7% for v1.5. GPT-Realtime-Translate handles 70+ input languages into 13 output languages in real-time. Zillow reported call success rates climbed 26 points using these models. Pricing: $32/1M audio input tokens, $64/1M output.
2026-04-17 / skill-finderOpenAI Codex CLI Ships GPT-5.3-Codex-Spark at 1,000+ Tokens/Sec, Windows OS-Level Sandboxing, and Realtime Voice v2 WebRTC DefaultOpenAI Codex CLI's April updates include a purpose-built GPT-5.3-Codex-Spark model running at 1,000+ tokens per second for rapid iteration loops, native Windows sandbox with OS-level networking isolation, and realtime voice sessions defaulting to the v2 WebRTC path with configurable transport and voice selection. MCP support expanded with resource reads, tool-call metadata, custom-server tool search, server-driven elicitations, and file-parameter uploads. The /resume command can now jump directly to a session by ID from the TUI. Signal for builders: Codex CLI is becoming a full agent runtime, not just a coding assistant.

Graph relationships

USES
Zillow
Zillow reported call success rates increased 26 points using GPT-Realtime-2.
Source finding
RELEASED
OpenAI
OpenAI launched GPT-Realtime-2 voice model with GPT-5-class reasoning.
Source finding
BUILT BY
OpenAI
GPT-Realtime-2 is built by OpenAI
Source finding
USES
Zillow -> GPT-Realtime-2
Zillow reported call success rates increased 26 points using GPT-Realtime-2.
Source finding
RELEASED
OpenAI -> GPT-Realtime-2
OpenAI launched GPT-Realtime-2 voice model with GPT-5-class reasoning.
Source finding

Source trail

xAI OpenAI Blog OpenAI Simon Willison OpenAI OpenAI Developers — Codex Changelog

Graph sources

entity graphfindings textkg edgeskg entities

Fetching from the wire…