Codex Day 7: GPT-6.1 Sol default context in Codex and Work rises from 272k to 500k, about 80% more

Why it matters

【Key points】

At 23:45 UTC on Oct 11 (07:45 Beijing time Oct 12), OpenAI Codex lead Tibo (@thsottiaux) posted Day 7 of the 28-day run: GPT-6.1 Sol's default context in Codex and ChatGPT Work rises from 272k to 500k tokens (~80%) to cut compactions on long tasks. He called 500k the new sweet spot, said it will go higher later, and noted users can still override it, with a 1M maximum in the post.

This changes the default, not the model. The model page still lists a 1,050,000-token window and 128,000 max output; API prompts over 272k input still bill the full request at 2x input and 1.5x output. At patrol time the Codex changelog had not listed it. About 14.5 hours later the post had ~1.02M views, ~10.2K likes and ~1,280 replies.

【Verifiable】

Tibo post: https://x.com/thsottiaux/status/2109430343346581527

Model page: https://developers.openai.com/api/docs/models/gpt-6.1-sol

Codex changelog (not yet listed): https://developers.openai.com/codex/changelog

X discussion (~7h from announcement hour): ~734 posts, peak ~230/hour

【Our read】

1. Another default-experience bump in the Codex 28-day cadence, alongside e/748 Ultrafast and e/727. The aim is fewer compressions on long work, not a new 1M marketing claim.

2. Default and API pricing still misalign: Codex/Work now defaults past 272k while the API surcharge still starts at 272k. Some secondary posts and an X news card claim no long-context surcharge on subscription usage; the primary post does not say that, so we do not treat it as confirmed.

3. Tool-heavy agents gain most. Less compaction means fewer lost signatures and error lines; 500k also buys more headroom if new threads still open by burning a large share of the window.