OpenAI speeds up Codex steering and adds GPT-6.1 Sol Ultrafast
OpenAI is making Codex respond sooner to changes during a task and introducing faster output for GPT-6.1 Sol.
OpenAI
OpenAI announced faster steering for Codex on October 8, allowing the coding assistant to react sooner when someone revises an ongoing task. The company also added an Ultrafast option for GPT-6.1 Sol in its Responses API.
Steering lets users correct instructions, supply additional context or redirect work already underway. The update targets the delay before Codex responds to those changes. Ultrafast addresses output speed by shortening the interval between generated tokens.
The steering rollout covers Codex within the ChatGPT desktop application. Its settings let users choose whether a follow-up changes the active task or waits for a subsequent run. API access to Ultrafast remains subject to rate limits, with processing options that include US and EU data residency.
The day's biggest AI news in one short email, every morning. Plus an instant heads-up when a reset or a big release lands.
More on OpenAI
- Oct 8OpenAI revenue reportedly nears $50 billion annual pace, below earlier estimatesNews
- Oct 8OpenClaw patch adds model support and fixes missing iMessage repliesOfficial
- Oct 9OpenAI's Tibo posts ChatGPT announcement without new product detailsOfficial
- Oct 8Meanwhile raises $37.5 million at a reported $350 million valuationNews
- Oct 8OpenAI bans Russian and Iranian operations using fake journalists and a think tankOfficial
- Oct 7Introducing GPT-6 in ChatGPT with Intelligent UIOfficial video