On this page

Usage analysis on reviewed Codex/Work accounts now attributes traffic to the model ID gpt-6.2-astra. That string appears beside older GPT-6 family labels such as gpt-6-astra and gpt-6.1-sol in the same dashboards.

What the usage charts show

On one account, a week spanning roughly 2026-10-26–11-01 puts gpt-6.2-astra near 6.6% of recorded usage; selecting November 1 alone shows gpt-6.2-astra at 100% for that day’s bar (no mix of the blue gpt-6-astra segment on that day). A second account shows the same ID around 22% for its window. Circulating X screenshots from other accounts show a similar label—one example near 37.7%—as social-media corroboration, not an OpenAI announcement. Selector UI in many sessions still reads GPT-6 Astra; a backend usage ID is not the same as a public listing.

Site-owner Codex/Work usage screenshot, week of 2026-10-26–11-01: gpt-6.2-astra ≈6.6%.
Site-owner Codex/Work usage screenshot, week of 2026-10-26–11-01: gpt-6.2-astra ≈6.6%.

Site-owner Codex/Work usage screenshot, week of 2026-10-26–11-01: gpt-6.2-astra ≈6.6%.

Same account usage screenshot, Nov 1 selected: gpt-6.2-astra = 100% for that day.
Same account usage screenshot, Nov 1 selected: gpt-6.2-astra = 100% for that day.

Same account usage screenshot, Nov 1 selected: gpt-6.2-astra = 100% for that day.

Circulating X usage screenshot (social corroboration; not the site-owner; not an official model page), one example near 37.7%.
Circulating X usage screenshot (social corroboration; not the site-owner; not an official model page), one example near 37.7%.

Circulating X usage screenshot (social corroboration; not the site-owner; not an official model page), one example near 37.7%.

Codex caps in ChatGPT plans: OpenAI Still Publishes No Weekly Codex Caps, So Here's What Each ChatGPT Plan Gets in October 2026. Broader token pricing context: OpenAI API Pricing (October 2026).

What official docs still omit

OpenAI’s public model catalog still surfaces family entries such as gpt-6-astra, gpt-6.1-sol, gpt-6.1-terra, and gpt-6-luna. Direct opens of /api/docs/models/gpt-6.2-astra and /api/docs/models/gpt-6-2-astra return 404. The recent API changelog does not carry a “released gpt-6.2-astra/GPT-6.2 Astra” style listing. A nearby Sol pricing/mode note: OpenAI adds Ultrafast mode to GPT-6.1 Sol: up to 8x faster, at six times the price. Codex iteration cadence context: Tibo announces 28-day plan: daily practical Codex improvements or a full reset.

What this means

A new label on a usage chart is not the same as an official model page. Until the public catalog, changelog, or pricing table lists it, treat gpt-6.2-astra as an account-side attribution—not a confirmed launch date. The percentages above are from the dashboards we checked; they do not speak for all Codex users or an official rollout calendar.

For reference only.