A model labelled gpt-6.2-astra appears in Codex usage screenshots posted on X on 1 November. In one, a usage breakdown lists gpt-6.2-astra at 19.8%, next to gpt-6-astra at 80% and gpt-6-luna at under 0.1%.
OpenAI has not announced a GPT-6.2 model. Our checks of Codex’s public model catalog, its recent releases, OpenRouter’s model list and OpenAI’s news feed found no reference to it.
Short explainer · GPT-6.2 Astra screenshots
A vertical news Short summarising this story. It reports posts on X and our own checks; the screenshots could not be verified.
Business Pill · THE MODEL AND THE HARNESS
A short explainer on models and harnesses. It teaches the general idea only and says nothing about GPT-6.2.
The key insight: As we read the posts, the only trace of gpt-6.2-astra we found on 1 November was two screenshots on X, one usage breakdown at 19.8% and one JSON snippet; our checks of Codex’s catalog, its releases, OpenRouter and OpenAI’s news feed found no reference.
At 10:09 UTC, the X account @TokenGremlin wrote that “OpenAI is quietly A/B testing a new variant called GPT-6.2 Astra, and it’s already showing up in Codex usage analytics”. It added that “The model picker still says GPT-6 Astra, but the usage breakdown for some accounts now lists gpt-6.2-astra separately”.
The account said “Multiple users are seeing the exact same thing” and called it “a gray rollout”. Its post did not name the users or say how many accounts were affected.
At 10:53 UTC, @kimmonismus posted “Confirmed. GPT-6.2 very close.” with a screenshot of a JSON snippet reading “response.completed” with the model field “gpt-6.2-astra”. The post did not say where the snippet came from.
Left: the usage split shown in a screenshot @TokenGremlin posted on X on 1 November 2026, which we could not verify. Right: the model slugs in Codex’s public catalog on GitHub, checked on 1 November 2026, which contains no 6.2 string.
What Was Posted on X
OpenAI is quietly A/B testing a new variant called GPT-6.2 Astra, and it’s already showing up in Codex usage analytics.
The model picker still says GPT-6 Astra, but the usage breakdown for some accounts now lists gpt-6.2-astra separately. Multiple users are seeing the exact same thing, so this isn’t a glitch. It’s a gray rollout, and a new Astra is coming soon.
@TokenGremlin posted at 10:09 UTC on 1 November that gpt-6.2-astra is showing up in Codex usage analytics while the model picker still says GPT-6 Astra, with a usage screenshot listing gpt-6.2-astra at 19.8%. We could not verify the screenshot.
GPT-6.2 Astra is already showing up in codex usage analytics, as per Token Gremlin.
Against I expect a release very soon, in response to Fable 5.5 and Anthropicโs IPO. Next week is going to be exciting.
@kimmonismus quoted that post at 10:51 UTC, citing Token Gremlin and writing that the account expects a release very soon. This piece reports the claim, not a confirmed release.
Confirmed. GPT-6.2 very close. Expect a release soon. Probably their new โbelโ flagship in response to Fable 5.5
At 10:53 UTC @kimmonismus posted ‘Confirmed. GPT-6.2 very close.’ with a JSON snippet whose model field reads gpt-6.2-astra. The post does not say where the snippet came from, and OpenAI has not announced the model.
What We Checked
Codex’s public model catalog on GitHub lists five model slugs: gpt-6-astra, gpt-6-luna, gpt-6-sol, gpt-6.1-sol and gpt-6.1-terra. The file contains no 6.2 string. Its latest change, on 26 October, added GPT-6.1 Terra.
The ten most recent Codex releases, the latest dated 29 October, do not mention 6.2 or Astra in their notes. OpenRouter lists GPT-6 Astra and GPT-6.1 models but no GPT-6.2, and OpenAI’s news feed, last updated 30 October, has no 6.2 item.
The Business Lever: Model Cadence Inside a Subscription
As we read the posts, the claimed rollout surfaced in a usage breakdown rather than in a launch post, and that breakdown is where Codex users see how their usage splits across models. OpenAI has said ChatGPT and Codex will share one usage allowance from December.
The Structural Read
As we read the timing, the catalog’s latest change, on 26 October, added GPT-6.1 Terra, so the newest modelOpenAI has put in that file is six days old.
As we read the claim, @TokenGremlin describes a test in which the model picker shows one name and the usage breakdown another; neither post shows that for more than the accounts in the screenshots.
As we read the line-up, OpenAI already sells several GPT-6 variants at different prices, so a new version label in a usage meter would sit inside that range.
@TokenGremlin on X, 1 November 2026
“The model picker still says GPT-6 Astra, but the usage breakdown for some accounts now lists gpt-6.2-astra separately.”
Three Implications
FOR CODEX USERS @TokenGremlin says the usage breakdown, not the model picker, is where gpt-6.2-astra appeared.
FOR OPENAI WATCHERS Codex’s public catalog lists gpt-6-astra, gpt-6-luna, gpt-6-sol, gpt-6.1-sol and gpt-6.1-terra, and no 6.2 model.
FOR RIVAL LABS @kimmonismus linked the timing to Fable 5.5, writing that GPT-6.2 is “Probably their new “bel” flagship in response to Fable 5.5”.
The analysis puts it in one line: “The surface, not the model, governs autonomy.”
As we read it, a model swap that shows up in Codex’s usage breakdown while the picker keeps the old name is the surface deciding which model runs. The framework was published in July 2026 and does not discuss GPT-6.2.
We could not verify the screenshots, the accounts they came from, or whether gpt-6.2-astra is an OpenAImodel name, a test label or something else. The second post also says GPT-6.2 is “Probably their new “bel” flagship in response to Fable 5.5”; it does not say what “bel” refers to. The 19.8% share is as shown in one screenshot.
This Week In The AI Supercycle
Anthropic’s Reality Check, The Capital Squeeze, & The New World Order
This week’s issue: for two years the AI build-out was priced as a growthstory. This week it started being priced as a credit story.
Screenshots posted on X on 1 November show gpt-6.2-astra in Codex usage data, at 19.8% of one breakdown. OpenAI has not announced the model, and Codex’s public catalog does not list it.
A note on sourcing. The screenshots and claims come from two X accounts and could not be verified. The catalog, release, OpenRouter and news-feed checks are our own, made on 1 November 2026 at about 11:20 UTC. We did not contact OpenAI.
Gennaro Cuofano is a CRO and tech executive who has worked in AI since late 2015, starting with bringing NLP, natural-language generation, voice and chatbot products to the marketing industry. His background is in law and finance: he holds a Master's degree in Law and an International MBA with an emphasis on corporate finance (LUISS Business School and the University of San Diego, 2012), and worked as a financial analyst at a real-estate investment firm in San Diego and as an assistant controller. His work focuses on business model strategy, business engineering and, more broadly, structural analysis: how companies make money, read from their own filings. He created FourWeekMBA and leads research there and at The Business Engineer, his newsletter on AI and business strategy, with over 95,000 subscribers and more than 1,000 published analyses. His writing has also appeared on Entrepreneur (for example, How to Digitize a Family Business in Europe) and HackerNoon. Find him on LinkedIn and Substack. See how we source.
