OpenAI to Launch GPT-5.6 Model with 1.5 Million‑Token Context and Bidirectional Audio

OpenAI is preparing to roll out the next generation of its language models, dubbed GPT‑5.6, as early as the week of 23 June 2026. The release may be staged across ChatGPT, Pro subscriptions, Codex and the API, with several variants – Mini, Standard and Pro – planned.

According to test leaks, GPT‑5.6 would expand the context window from 1 million to up to 1.5 million tokens, enabling it to handle very large codebases, long documents and complex, multi‑step tasks in a single session. The model is also reported to improve reasoning, coding assistance and agent‑based workflows, and to introduce a bidirectional audio system (GPT‑Bidi‑1) that can listen and speak simultaneously, offering more natural conversational interaction. No official launch date or detailed specifications have been confirmed by OpenAI.