Key takeaways

  • As of November 18, 2026, OpenAI has not announced a GPT-6.2 Luna or given any release date for one.
  • The current Luna is GPT-6 Luna (gpt-6-luna), released September 22, 2026, at $0.10 per million input tokens and $0.50 per million output tokens.
  • GPT-6 Luna has a 1,050,000-token context window, 128,000 output tokens, and powers OpenAI's Decisions API.

OpenAI has not announced a GPT-6.2 Luna, and there is no release date for one. As of November 18, 2026, the Luna model in OpenAI's API model list is still GPT-6 Luna (gpt-6-luna), released on September 22 at $0.10 per million input tokens and $0.50 per million output tokens.

The search makes sense. On November 17, OpenAI released GPT-6.2 Astra and GPT-6.2 Sol, and people who use the cheapest model in the family want to know when their turn comes. OpenAI's own pages don't say.

Is there a GPT-6.2 Luna release date

No. Here is what OpenAI's pages say, checked on November 18:

  • The API changelog entry for November 17 announces GPT-6.2 Astra and GPT-6.2 Sol only. No entry in October or November mentions an upcoming Luna.
  • The same day, a second changelog entry added cheaper background completion windows for four models: gpt-6.2-astra, gpt-6.2-sol, gpt-6.1-terra and gpt-6-luna. The newest Luna OpenAI names there is the GPT-6 one.
  • The models page lists four main models: GPT-6.2 Astra, GPT-6.2 Sol, GPT-6.1 Terra and GPT-6 Luna. There is no gpt-6.2-luna.
  • The GPT-6.2 Astra and Sol deployment safety page covers those two models and doesn't mention Luna.
  • The ChatGPT and Codex changelog says GPT-6.2 Astra and Sol came to ChatGPT Work and Codex on November 17. It doesn't mention Luna.

So "GPT-6.2 Luna" is a name people are guessing at, not a product. Any date you see for it elsewhere does not come from OpenAI.

What GPT-6 Luna is for

OpenAI calls GPT-6 Luna "our most efficient model for focused, high-volume tasks." Think of jobs where you send the same kind of request thousands or millions of times: sorting support tickets, pulling fields out of documents, tagging images, routing requests to a bigger model.

It is a reasoning model, which means it can spend extra tokens thinking before it answers. You choose how much with the reasoning.effort setting, from none up to max, with medium as the default. It reads text and images and writes text. It works through the Responses, Chat Completions and Batch endpoints, but not the Realtime API, fine-tuning or embeddings.

Luna is also the model behind OpenAI's Decisions API, which turns text and images into typed answers, such as a yes or no or a pick from a list. The API entered beta with gpt-6-luna on October 6, and OpenAI claimed it is 10 times faster than the Responses API for this kind of work. It became generally available on October 22.

What GPT-6 Luna costs

Prices per million tokens, from OpenAI's model page, as of November 18, 2026:

GPT-6 Luna Price per 1M tokens
Input $0.10
Cached input $0.01
Cache writes $0.125
Output $0.50

A few rules change the bill:

  • Prompts over 272,000 input tokens cost 2x for input and cache, and 1.5x for output, on the whole request.
  • Batch and Flex processing cost half the standard rate. Fast mode costs double.
  • The new background completion windows from November 17 cut token rates by 30% for a two-hour window and 40% for an eight-hour window, compared with standard.
  • Regional processing adds 10% where it is offered.

The context window is 1,050,000 tokens, with up to 922,000 tokens of input and 128,000 tokens of output. The knowledge cutoff is May 18, 2026.

How Luna compares with the GPT-6.2 models

Luna is far cheaper than anything else in the main lineup. Standard prices per million tokens on OpenAI's models page, as of November 18, 2026:

Model Input Output Context
GPT-6.2 Astra $8 $40 1.05M
GPT-6.2 Sol $2 $10 1.05M
GPT-6.1 Terra $1 $5 1.05M
GPT-6 Luna $0.10 $0.50 1.05M

Sol costs 20 times as much as Luna per input token, and Astra 80 times. All four share the same context window and 128,000-token output limit. For the differences between the new models and the ones they replaced, see GPT-6.2 vs 6.1.

OpenAI has updated its models one tier at a time

OpenAI's changelog shows the GPT-6 family has not moved in lockstep:

Date (2026) Release
September 3 GPT-6 Astra
September 22 GPT-6 Sol and GPT-6 Luna
September 29 GPT-6.1 Sol
October 26 GPT-6.1 Terra
November 17 GPT-6.2 Astra and GPT-6.2 Sol

There was never a GPT-6.1 Luna. Luna went from GPT-6 straight past the 6.1 releases, and it has now been passed over by 6.2 too. The previous generation was different: on July 9, OpenAI released GPT-5.6 Sol, Terra and Luna together as one family.

That history doesn't predict anything. It only shows that a new version number for Astra or Sol has not meant a new Luna in the GPT-6 era, so there is no reason to hold off on GPT-6 Luna while waiting for a 6.2 version.

What to watch

If OpenAI ships a new Luna, it will show up first in the API changelog and on the models page, with a model ID like the others. Until then, gpt-6-luna is the current Luna, and its model page lists a single snapshot under that same ID.

Frequently asked questions

When is GPT-6.2 Luna coming out?

OpenAI has not announced a GPT-6.2 Luna or a date for one. Its November 17, 2026 release covered only GPT-6.2 Astra and GPT-6.2 Sol.

What is the latest Luna model?

GPT-6 Luna, model ID gpt-6-luna, released in the API on September 22, 2026. OpenAI's models page lists no newer Luna as of November 18, 2026.

Is GPT-6 Luna free?

No, not in the API. It costs $0.10 per million input tokens and $0.50 per million output tokens, with Batch and Flex at half that.

GPT-6 Luna vs GPT-6.2 Sol: which should I use?

Luna is for high-volume, focused tasks and costs a twentieth of Sol's $2 input price. OpenAI positions Sol for everyday coding and agent work.

Sources

  1. API changelogOpenAI
  2. GPT-6 Luna model pageOpenAI
  3. ModelsOpenAI
  4. GPT-6.2 Astra and GPT-6.2 Sol deployment safetyOpenAI
  5. ChatGPT and Codex changelogOpenAI

Add ClaudeGPT Si as a preferred source on Google