ChatGPT 6 is here: GPT-6 Astra, GPT-6 Pro, access, context, API pricing, and what changed
Updated: Sep 17
GPT-6 is no longer a future model. OpenAI released GPT-6 Astra on September 3, 2026, and ChatGPT now exposes it as GPT-6 Pro on eligible plans while the API uses the model ID gpt-6-astra.
Astra is positioned above the GPT-5.6 family for the hardest end-to-end work across reasoning, coding, computer use, browsing, science, cybersecurity, and professional workflows.
[[ADPLUS_300x250_1]]
··········
GPT-6 ASTRA LAUNCHED ON SEPTEMBER 3, 2026
The original version of this page tracked what was known before release. That lifecycle is now complete: GPT-6 Astra is a shipping model, not a rumor or roadmap placeholder.
OpenAI describes Astra as its most capable broadly deployed model and its new flagship for difficult multistep work.
........
Dimension | GPT-6 Astra |
|---|---|
API model ID | gpt-6-astra |
Context window | 1,050,000 tokens |
Maximum output | 128,000 tokens |
Knowledge cutoff | April 30, 2026 |
Reasoning effort | low, medium, high, xhigh, max |
API input price | $10 per 1M tokens |
Cached input | $1 per 1M tokens |
Cache writes | $12.50 per 1M tokens |
API output price | $50 per 1M tokens |
........
··········
GPT-6 PRO IS THE CHATGPT NAME, GPT-6 ASTRA IS THE MODEL NAME
In standard ChatGPT, Astra appears as GPT-6 Pro for eligible Pro, Business, and Enterprise users. Enterprise availability can also depend on workspace model permissions.
Plus does not receive GPT-6 Pro in ordinary Chat according to the current ChatGPT access documentation, but Plus includes GPT-6 Astra in ChatGPT Work and Codex.
··········
THE API EXPOSES GPT-6 ASTRA DIRECTLY
Developers use gpt-6-astra through the OpenAI API. Astra supports text and image input, text output, a 1.05-million-token context window, and up to 128,000 output tokens.
Requests above 272,000 input tokens use higher long-context pricing: input and cache rates are doubled and output pricing is multiplied by 1.5 for the full request.
··········
ASTRA ADDS ASYNC TOOL CALLING AND MID-TURN STEERING
GPT-6 Astra can continue reasoning while an application executes a tool asynchronously. Developers can also send new instructions while the model is working, allowing a running task to absorb corrections or changed requirements without restarting the workflow from zero.
Astra also supports changing reasoning effort during a conversation while preserving prompt-cache continuity, alongside computer use, structured outputs, streaming, programmatic tool calling, multi-agent orchestration, prompt caching, persisted reasoning, and compaction.
··········
GPT-6 PRO IS NOT UNLIMITED IN CHATGPT
GPT-6 Pro has plan-specific message allowances in ChatGPT. The Pro $100 tier currently shares a 50-message weekly allowance across GPT-6 Pro and GPT-5.6 Sol Pro, while higher tiers use different allowances. Reaching the Pro-model limit does not remove the subscription; the user can switch to another available model or wait for the allowance to reset.
··········
GPT-6 ASTRA SITS ABOVE GPT-5.6, BUT THE TWO FAMILIES COEXIST
GPT-5.6 Sol, Terra, and Luna remain important because they cover different capability, latency, and cost points. Astra is not simply a universal replacement for every request: its API price is higher, and ChatGPT access is more constrained than everyday GPT-5.6 use.
A useful architecture is therefore Astra for the hardest end-to-end work, Sol for advanced professional reasoning, Terra for balanced workloads, and Luna for high-volume low-cost tasks.
··········
CYBERSECURITY CAPABILITY COMES WITH STRONGER SAFEGUARDS
OpenAI classifies GPT-6 Astra at the Critical cybersecurity capability level under its Preparedness Framework. The company says it strengthened isolation, monitoring, checkpoint protection, misuse controls, and alignment evaluation before deployment.
Those safeguards can affect how Astra handles high-risk cybersecurity requests even when the underlying model is technically capable of more.
··········
WHAT CHANGED FROM THE PRE-RELEASE CHATGPT-6 STORY
Before September 2026, the useful question was whether a GPT-6-class model was coming and what signals pointed toward it. The current question is operational: where Astra is available, how GPT-6 Pro differs from GPT-5.6 Pro, what the API costs, and when the extra capability justifies the higher price and tighter allowance.
··········
FOLLOW US FOR MORE
··········
DATA STUDIOS
··········
