Expired50% confidenceReleaseExact time
OpenAI launched GPT Real-Time 2.1 and its Mini version real-time voice models, made available via API, with a new caching mechanism that can reduce P95 latency by at least 25%.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
7/20/2026
First Seen
Valid until: 10/18/2026(expired)
Sources
Related Claims
VerifiedGPT-Realtime-2 is a real-time voice interaction model launched by OpenAI.84% similarUnverifiedOpenAI released GPT Live, supporting full-duplex voice and interruptible interaction, with latency reduced to real-time levels and weekly active users reaching 150 million.82% similarVerifiedGPT-4o released by OpenAI in 2024 achieved low-latency real-time voice conversation, with end-to-end latency reduced to under 320 milliseconds.80% similarUnverifiedOpenAI launched the GPT 5.3 Instant model, with the key improvement being a reduction in sycophantic conversational style, making conversations more natural and fluid.78% similarVerifiedOpenAI released a new-generation voice model, GPT-Live, and simultaneously launched it in ChatGPT.78% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/570052API
curl https://kongchang.com/api/v1/knowledge/claims/570052MCP
get_claim(id=570052)