{
"$type": "site.standard.document",
"bskyPostRef": {
"cid": "bafyreicotdb65zzefirqic6zibcez6pqfjdwax6twm47bciinelpxrfhyy",
"uri": "at://did:plc:lk3jfj3zq4k4wxnk474axylu/app.bsky.feed.post/3mfkwi7t7pfu2"
},
"path": "/t/gpt-realtime-1-5-is-live-in-realtime-api/1374919#post_1",
"publishedAt": "2026-02-23T23:43:09.000Z",
"site": "https://community.openai.com",
"tags": [
"Realtime API | OpenAI API"
],
"textContent": "Hey everyone,\n\nBig step forward for voice today: `gpt-realtime-1.5` just dropped in the Realtime API.\n\nQuick highlights from the team:\n\n * +5% on Big Bench Audio reasoning\n * +10.23% alphanumeric transcription accuracy\n * +7% instruction following\n * More reliable tool calling and multilingual handling overall\n\n\n\nPricing remains steady with the original gpt-realtime:\n\n * **Text:** $4 / 1M input | $0.40 cached | $16 / 1M output\n * **Audio:** $32 / 1M input | $0.40 cached | $64 / 1M output\n\n\n\nEarly adopters are already feeling the upgrade:\n\n * **Genspark** reports connection rates nearly doubled (up to **66%** ) and phone call errors cut in half.\n * **Sendbird** highlights exceptional improvements in handling interruptions.\n\n\n\nCheck out the latest docs here: Realtime API | OpenAI API\n\nCurious about your experience:\n\n * Are you noticing reduced latency in your setups?\n * Any standout improvements or quirks in tool calling and multilingual tasks?\n * How does it stack up side-by-side with previous realtime models?\n\n\n\nDrop your insights, benchmarks, or any questions right here!\n\nExcited to hear your thoughts!",
"title": "🚀 gpt-realtime-1.5 is live in Realtime API"
}