GPT-6 Release Date: What Is Actually Confirmed?

OpenAI has not officially confirmed a model called “GPT-6” as its next flagship model.
The most defensible current conclusion is that the flagship model family is the one OpenAI has publicly documented under the name “GPT-6 Astra” in its own product and developer pages, but the exact branding across products is still easy to confuse because OpenAI separates:

1. announcement
2. ChatGPT availability
3. API availability

So the safest wording is:

• Name: “GPT-6” / “GPT-6 Astra” appears in OpenAI’s own materials, but the precise flagship branding is still a product-level naming issue, not a single universally confirmed one.
• Release timing: OpenAI has already announced and rolled out the model family in stages.
• Variants: Astra is the flagship; Sol and Luna are lower-cost/faster variants; a later GPT-6.1 Sol update is also documented.
• Benchmarks: OpenAI publishes strong benchmark claims, but they are OpenAI-reported and not independently verified in the documentation itself.
• Pricing: OpenAI has published API pricing for the model family, with higher prices for the flagship and much lower prices for Sol/Luna.

What is confirmed

1) Model name

OpenAI’s own pages reference GPT-6 Astra and related models such as GPT-6 Sol, GPT-6 Luna, and GPT-6.1 Sol. That is the strongest public evidence that the next flagship generation is being marketed under the GPT-6 umbrella.

What is not confirmed from the documentation is a clean, standalone “this model will simply be called GPT-6” statement in the way people usually mean a final product name. The confirmed naming is more specific and productized.

2) Release timing

The evidence points to a staged rollout, not one single launch date:

• Announcement: OpenAI first introduced the flagship model publicly.
• ChatGPT availability: access expanded later and depends on plan/workspace.
• API availability: developer access is documented separately, with model IDs and pricing.

So if you’re asking for one exact date, the answer is: there isn’t one single “release date” that covers announcement, ChatGPT, and API at once.

3) Model variants

The confirmed variants are:

• GPT-6 Astra — flagship / highest capability
• GPT-6 Sol — lower-cost, strong general/professional model
• GPT-6 Luna — fastest / cheapest high-volume model
• GPT-6.1 Sol — later refresh of Sol

4) Benchmarks

OpenAI has published benchmark claims for the flagship and related models, but these should be treated as vendor-reported results unless independently reproduced.

The documentation includes claims about performance on areas like:

• reasoning / mathematics
• coding and software engineering
• computer use / agentic workflows
• cybersecurity-related evaluations
• terminal / scientific tasks

Important caveat: benchmark numbers are not the same as independent confirmation. They are evidence of OpenAI’s claims, not external validation.

5) Pricing

OpenAI has published API pricing for the family. The general shape is:

• Astra: premium flagship pricing
• Sol: mid-tier pricing
• Luna: very low-cost pricing

The exact quoted rates vary by model and request type, and the flagship also has special rules for long-context or fast-mode usage. ChatGPT access is mostly subscription/plan-based, not a simple per-model retail price.

What is still unconfirmed or should be treated cautiously

• A single universal “GPT-6” launch date
• Parameter count
• Architecture details
• Independent benchmark validation
• Whether every branded variant is available in every product surface in every region

Bottom line

The current confirmed evidence supports this summary:

OpenAI’s next flagship generation is publicly documented under the GPT-6 / GPT-6 Astra naming umbrella, with Sol and Luna as lower-cost variants and GPT-6.1 Sol as a later refresh. OpenAI has already published staged release, ChatGPT, and API details, plus pricing and benchmark claims—but the benchmark numbers remain OpenAI-reported, and a single, clean “GPT-6 release date” is not the right way to describe the rollout.