OpenAI GPT Model Release Timeline: ChatGPT to GPT-6 Astra and GPT-6.1 Sol (2022–2026)
Live timeline · last updated · 1 entry in the update log
One dated list of OpenAI’s GPT and o-series models, from the ChatGPT research preview to the GPT-6 family. API dates come from OpenAI’s own changelog; the GPT-6 entries come from openai.com and the Help Center release notes.
As of Oct 21, 2026 OpenAI’s newest models are GPT-6 Astra (Sep 3, 2026), GPT-6 Sol and Luna (Sep 22) and GPT-6.1 Sol (Sep 29), with an updated GPT-6 Pro announced in ChatGPT on Oct 20; the line started with ChatGPT on Nov 30, 2022.
Read on 2026-10-21: the OpenAI API changelog on developers.openai.com, the openai.com announcement for ChatGPT, GPT-4, GPT-5.4, the path to Astra, GPT-6 Astra, GPT-6 Sol and Luna and GPT-6.1 Sol, and the ChatGPT and model release notes on help.openai.com. Where a model reached ChatGPT and the API on different days, we use the API changelog date unless an openai.com post gives one launch date. We ran no benchmarks.
Timeline
-
OpenAI launches ChatGPT as a free research preview, fine-tuned from a model in the GPT-3.5 series.
-
GPT-4 is announced as a large multimodal model that accepts image and text inputs, available to ChatGPT Plus users and through an API waitlist.
-
GPT-4 Turbo Preview is released in the API, alongside the Assistants API and DALL·E 3 in the API.
-
GPT-4 Turbo with Vision becomes generally available in the API.
-
GPT-4o is released in the API; the changelog calls it OpenAI’s fastest and most affordable flagship model at the time.
-
GPT-4o mini is released as a small, low-cost model.
-
o1-preview and o1-mini launch, the first o-series models, trained with reinforcement learning to perform complex reasoning tasks.
-
o3-mini is launched, a small reasoning model the changelog says is optimized for science, math and coding tasks.
-
A research preview of GPT-4.5 is released, which the changelog calls OpenAI’s largest and most capable chat model at the time.
-
GPT-4.1, GPT-4.1 mini and GPT-4.1 nano are released in the API with context windows of up to 1M tokens.
-
o3 and o4-mini are released as the next o-series reasoning models.
-
o3-pro is released in the API for harder reasoning problems.
-
The GPT-5 family is released in the API: gpt-5, gpt-5-mini and gpt-5-nano, with a new minimal reasoning-effort setting.
-
GPT-5 Pro becomes available in the API.
-
GPT-5.1 is released in the API.
-
GPT-5.2 is released, with the changelog listing gains over GPT-5.1 in general intelligence, instruction following, vision, front-end code and spreadsheets.
-
gpt-5.3-codex, a coding-specialized model, is released to the Responses API.
-
GPT-5.3 Instant reaches the API as gpt-5.3-chat-latest, the chat model used in ChatGPT.
-
GPT-5.4 and GPT-5.4 Pro are released.
-
GPT-5.4 mini and GPT-5.4 nano are released in the API.
-
GPT-5.5 and GPT-5.5 Pro are released in the API.
-
The GPT-5.6 family is released in three sizes named Sol, Terra and Luna, the naming OpenAI carries into GPT-6.
-
OpenAI publishes “The path to Astra”, saying its next model is the first it treats as reaching the Critical cybersecurity capability level under its Preparedness Framework, and describing a staged rollout.
-
GPT-6 Astra is released at $10 / $50 per million input and output tokens. It goes first to a limited set of organizations, then to ChatGPT Plus, Pro, Business and Enterprise, the API, Microsoft Azure and Amazon Bedrock.
-
GPT-6 Sol and GPT-6 Luna are released. OpenAI prices Sol at $2 / $10 and Luna at $0.10 / $0.50 per million tokens, 50% below the GPT-5.6 equivalents.
-
GPT-6.1 Sol is released at $2 input, $0.10 cached input and $10 output per million tokens. The same day the changelog adds an Ultrafast mode for GPT-6 Astra in the API.
-
The ChatGPT release notes announce GPT-6 with Intelligent UI in ChatGPT.
-
Ultrafast mode is added for GPT-6.1 Sol in the Responses API, available to all API users subject to rate limits.
-
GPT-5.5 is retired from ChatGPT, according to OpenAI’s model release notes.
-
The ChatGPT release notes announce an updated GPT-6 Pro that can run parallel agents.
How the dates were chosen
OpenAI often launches a model in ChatGPT, the API and partner clouds on different days, and its own pages do not always agree on a single date. To keep this list consistent we use one rule: for API models, the date in the OpenAI API changelog; for launches with a dedicated openai.com post, the date of that post; for ChatGPT-only changes, the date of the Help Center release notes entry. One discrepancy is worth naming. The ChatGPT release notes list “Introducing GPT-6 Astra” under a September 11, 2026 heading, but OpenAI’s GPT-6 Astra post and the API changelog both point to September 3, so we use September 3.
We list model generations and named variants, not every dated snapshot, audio or image model, or fine-tuning feature. Open-weight models such as gpt-oss and the image, video and voice models deserve their own timelines.
The naming, in one paragraph
OpenAI ran two lines in parallel for about a year. The GPT line (GPT-4, GPT-4o, GPT-4.1, GPT-4.5) was for general chat, and the o-series (o1, o3, o4-mini, o3-pro) was for reasoning. GPT-5 in August 2025 merged them into one family with mini and nano sizes and a Pro tier. In July 2026 GPT-5.6 replaced mini and nano with three named sizes, Sol, Terra and Luna. GPT-6 kept Sol and Luna and added Astra above them, a model OpenAI handled differently because of its cybersecurity capability.
2026 in detail: from GPT-5.4 to GPT-6
OpenAI shipped five numbered generations in 2026 before October: GPT-5.3 (codex in February, Instant in March), GPT-5.4 on March 5, GPT-5.5 on April 24, GPT-5.6 on July 9 and GPT-6 in September. GPT-5.4 to GPT-5.5 took seven weeks, while GPT-6.1 Sol followed GPT-6 Astra by just 26 days.
GPT-6 arrived in two steps. On September 1 OpenAI published “The path to Astra”, saying the model reached the Critical level for cybersecurity under its Preparedness Framework, which is why access was staged. Two days later, on September 3, GPT-6 Astra launched at $10 / $50 per million tokens, first for a limited set of organizations and then for paid ChatGPT plans, the API, Azure and Bedrock. The cheaper GPT-6 Sol and GPT-6 Luna followed on September 22, and GPT-6.1 Sol a week later on September 29.
The October entries are mostly product changes on top of those models: GPT-6 with Intelligent UI in ChatGPT on October 7, Ultrafast mode for GPT-6.1 Sol on October 8, GPT-5.5 leaving ChatGPT on October 14, and an updated GPT-6 Pro with parallel agents on October 20.
Which OpenAI model is the newest, by tier
- Most capable: GPT-6 Astra, released September 3, 2026, at $10 / $50 per million tokens.
- Mid-size: GPT-6.1 Sol, released September 29, 2026, at $2 / $10, with $0.10 cached input.
- Smallest: GPT-6 Luna, released September 22, 2026, at $0.10 / $0.50.
- ChatGPT Pro tier: an updated GPT-6 Pro with parallel agents, announced October 20, 2026.
What is not verified here
- Release dates for GPT-6.1 Astra, GPT-6.1 Luna or GPT-7. OpenAI has not announced them, and we do not list rumored dates.
- Benchmark numbers. OpenAI’s pages publish many; we did not reproduce them and cite none as facts.
- Parameter counts and training details, which OpenAI has not published for these models.
- Exact ChatGPT rollout days by plan and country, which vary and are not all recorded in the release notes.
Related timelines
Compare the Claude model timeline, the Gemini model timeline and the Grok model timeline. For one OpenAI story covered in depth, read Terence Tao vs OpenAI. Every post is on the Front blog index.
The same format also covers Moonshot Kimi, Mistral, DeepSeek and Alibaba Qwen.
Keep reading on Front
Every post, newest first, is on the Front blog index.
- Claude model release timeline: every Claude version with its release date.
- Gemini model release timeline: every Gemini version from 1.0 to Gemini 4 Argon.
- Grok model release timeline: every Grok version, with Elon Musk’s posts quoted and linked.
- Terence Tao vs OpenAI: one mathematician’s dated reaction to the Oct 6 release.
- AI model releases in 2024: the 2024 nodes from the vendor timelines, in date order.
- AI model releases in 2025: the 2025 nodes from the vendor timelines, in date order.
- AI model releases in 2026: the 2026 nodes from the vendor timelines, in date order.
- Mistral model timeline: Mistral from 7B to Large 4, with the license on each release.
FAQ
What is OpenAI’s latest model?
As of October 21, 2026: GPT-6 Astra (September 3) is the most capable, GPT-6.1 Sol (September 29) is the newest mid-size model, and GPT-6 Luna (September 22) is the smallest. An updated GPT-6 Pro was announced for ChatGPT on October 20.
When was GPT-6 Astra released?
September 3, 2026, after OpenAI described its staged rollout in “The path to Astra” on September 1.
When was GPT-5 released?
The GPT-5 family (gpt-5, gpt-5-mini, gpt-5-nano) was released in the API on August 7, 2025.
When did ChatGPT launch?
November 30, 2022, as a free research preview.
Update log
- — First published. Checked openai.com, the OpenAI API changelog and the Help Center release notes through Oct 21, 2026 (UTC+8).