AI News
All stories
Rumor

Rumors say GPT-6.2 Astra is close, citing Codex usage analytics

Commentators say GPT-6.2 Astra is visible in Codex usage analytics and may release very soon, but the claim is unconfirmed.

Models · · 1 min read

Two posts on X say GPT-6.2 Astra is close to release. The evidence in them is a model name seen in usage data, not an announcement. Chubby (@kimmonismus) wrote on November 1, 2026 that GPT-6.2 was 'very close' and labeled it 'Confirmed,' predicting a release 'very soon,' probably as a flagship answering Fable 5.5. The post doesn't say who confirmed it or how.

The underlying observation comes from Token Gremlin (@TokenGremlin), who wrote that OpenAI is quietly A/B testing a variant called GPT-6.2 Astra. The model picker still reads GPT-6 Astra. However, the usage breakdown for some accounts lists gpt-6.2-astra as a separate entry, and Token Gremlin says multiple users see the same one. A separate model ID in a usage breakdown suggests that a backend model carries that name. It does not establish that the model will ship, when it will ship, or whether the entry is a test build that may never reach users.

Token Gremlin also said the finding may be the surprise that 'Tibo had planned alongside the two banked resets,' and that the launch is 'probably sometime this week.' Chubby's follow-up tied the Codex finding to Fable 5.5 and Anthropic's IPO and said next week 'is going to be exciting.' Both posts rest their timing on the authors' expectations. The linked pages contain no OpenAI statement and no release date.

A separate item in the same roundup says Sol 6.1 and Astra score very similarly on GDPx, a benchmark of 206 agentic tasks where the best model scores only 12%.

Why it matters

A near-term model release could change model choices for teams building on OpenAI, so procurement plans may need a flexible review point.

Original source

Further reporting

Where we saw it

1 source

Latent Space

AI wrote this summary from the sources listed, and nobody reviewed it individually before publication. Check the original source before you rely on it. How it works