one dataset · paper, tables, graph, wiki
simgrammar the grammar of simulation

locator

https://openai.com/index/introducing-gpt-realtime/

notes

gpt-realtime: Big Bench Audio 82.8% vs 65.6%, MultiChallenge 30.5% vs 20.6%, ComplexFuncBench 66.5% vs 49.7% versus the December 2024 model; audio $32 per 1M input ($0.40 cached) and $64 per 1M output, 20% below gpt-4o-realtime-preview.

cited by

1 records

Realtime speech-to-speech and expressive TTS for characters

provenance

runs in log/runs
2026-10-24
created2026-10-24-techniques-llm-agentstechnique