Short answer first, because that is what you came for: no, GPT-6 Astra is not free. It is not on the ChatGPT Free tier and it is not on the $8 Go tier. Beyond that the picture is genuinely more complicated than most articles admit, because access differs not just by plan but by which OpenAI product you open.
Direct Answer: GPT-6 Astra is not available on ChatGPT's Free or Go tiers. It shipped as a phased rollout in which access differed by plan and by product: Pro, Business and Enterprise plans received GPT-6 Pro (powered by Astra) inside Chat, while Plus subscribers initially got Astra through ChatGPT Work and Codex rather than the ordinary chat model picker. On subscription plans there is no per-message charge — you pay the monthly fee and work within a message cap. Via the API, pricing is entirely separate and usage-based: $10 per million input tokens, $1 per million cached input tokens, $12.50 per million cache-write tokens, and $50 per million output tokens. Because rollout details changed repeatedly during the first week, verify current plan availability on OpenAI's own pricing page before you buy.
The short answer by plan
| Plan | Approximate price | GPT-6 Astra access |
|---|---|---|
| ChatGPT Free | $0 | Not included |
| ChatGPT Go | $8 / month | Not included |
| ChatGPT Plus | $20 / month | Astra via ChatGPT Work and Codex during rollout; not in the standard chat model picker initially |
| ChatGPT Pro | Reported at $100 and $200 / month tiers | GPT-6 Pro (Astra-powered) in Chat, with a weekly message cap |
| ChatGPT Business | $25 / user / month monthly, $20 / user / month annual (Standard); $125 / $100 for Premium seats | Astra-powered access; Premium seats get the higher caps |
| ChatGPT Enterprise | Custom quote | Included |
| OpenAI API | Pay per token | Available as gpt-6-astra |
Two honest caveats about that table. First, the rollout was phased and moved several times in its first week — treat any third-party availability table, including this one, as a snapshot rather than a specification. Second, reported weekly caps for GPT-6 Pro messages differed by tier (higher on the more expensive Pro tier, lower on the cheaper one and on Business Premium seats). Because those figures come from reporting rather than a first-party specification page we can cite, check them in your own account's plan details rather than relying on a number you read in an article.
What "included" actually means
On a subscription, "included" does not mean unlimited. It means you pay a flat monthly fee and get a message allowance, typically expressed weekly for the highest-compute models. When you exhaust it you either wait for the window to reset or fall back to a lesser model.
This matters when comparing routes. A $200 monthly plan with a 200-message weekly allowance is roughly 800 messages a month — a real limit if you are working through the model all day, and generous if you use it a few times a week. The relevant comparison is not "subscription versus API" in the abstract; it is your actual message volume against your actual token volume.
API pricing is a completely separate thing
A ChatGPT subscription does not include API credits, and API spending does not count toward any subscription. They are separate products with separate billing.
| Token type | Price per million tokens |
|---|---|
| Input | $10.00 |
| Cached input | $1.00 |
| Cache write | $12.50 |
| Output | $50.00 |
The detail that catches people out is that output tokens cost five times input tokens. A verbose model configuration is not slightly more expensive than a terse one; it is meaningfully more expensive. Reasoning tokens are also billed at the output rate, so the reasoning-effort setting is a pricing decision.
Two discounts apply automatically or nearly so. Prompt caching bills repeated input at 0.1× the standard rate, and the Batch API takes 50% off in exchange for a 24-hour completion window.
What you can do without paying for Astra
- Use a cheaper model. GPT-5.6 Sol ($4 / $20), Terra ($2 / $12) and Luna ($0.20 / $1.20) all have the same 1,050,000-token context window and the same 128,000-token output cap as Astra. For most everyday work the difference is not the thing standing between you and your goal.
- Use ChatGPT Free or Go for interactive work on whichever models those tiers do include.
- Start with API credits rather than a subscription if your usage is occasional. A few dollars of tokens goes a long way at Luna or Terra rates.
Which route is cheaper for you?
Work it out rather than guessing. Estimate your monthly token volume, price it, and compare against the subscription.
Worked example — light use. Say 200 requests a month, each with roughly 2,000 input tokens and 800 output tokens:
- Input: 200 × 2,000 = 400,000 tokens → 0.4 × $10 = $4.00
- Output: 200 × 800 = 160,000 tokens → 0.16 × $50 = $8.00
- Total: $12.00/month — cheaper than a $20 Plus subscription, and you get API access rather than a chat interface.
Worked example — heavy interactive use. Say 40 substantial conversations a day, each accumulating 25,000 input tokens across turns and generating 4,000 output tokens, over 22 working days:
- Input: 40 × 22 × 25,000 = 22,000,000 tokens → 22 × $10 = $220
- Output: 40 × 22 × 4,000 = 3,520,000 tokens → 3.52 × $50 = $176
- Total: $396/month — at which point a Pro subscription looks like better value, if your usage fits inside its message cap.
The crossover point is real and it is not intuitive, which is the whole reason to do the arithmetic. Rather than working the examples above by hand for your own numbers, our AI Token Counter & API Cost Calculator takes a prompt, a model and a request volume and returns the monthly figure directly — which is the number to hold against a subscription price. If you want to work out what percentage a caching or model change would save off a current bill, our Percentage Calculator handles the increase-and-decrease direction correctly — a 90% cached-token discount and a 2.5× model premium do not cancel out the way people assume they do.
The heuristic underneath the examples: API billing scales with volume; subscriptions scale with nothing. Occasional or automated use favours the API. Constant interactive use favours a subscription, as long as you fit within the cap.
A note for readers outside the US
Every price above is in US dollars, and that is how OpenAI publishes them. Local pricing, taxes and payment-processor fees can change what actually lands on your card, and some plans are priced regionally.
For a current conversion rather than a rate from whenever an article was written, our Currency Converter will give you today's figure — worth doing before you commit to a recurring monthly charge in a foreign currency, since a $200 subscription is a very different number depending on the month.
Frequently Asked Questions (FAQs)
Is GPT-6 Astra free on ChatGPT?
No. Neither the Free tier nor the $8 Go tier includes it. You need at least Plus for the Work and Codex routes, or Pro, Business or Enterprise for Astra-powered access in Chat.
Is there a free trial for GPT-6 Astra?
OpenAI did not announce a dedicated Astra trial. The nearest equivalent is API access, where you only pay for what you use — a small number of test calls costs cents rather than a monthly commitment.
Why does GPT-6 Astra appear in Codex but not in my ChatGPT model picker?
That was the actual shape of the rollout for Plus subscribers: Astra became available inside ChatGPT Work and Codex before appearing in the ordinary chat model list. It is a phased rollout, not an account fault.
Does my ChatGPT Plus subscription include API access?
No. They are separate products with separate billing. Paying for Plus gives you nothing in the API, and API spending gives you nothing in ChatGPT.
What is the cheapest way to try GPT-6 Astra?
The API. A handful of test calls at $10 per million input and $50 per million output tokens costs a few cents, versus committing to a monthly subscription to find out whether it helps.
Will Astra get cheaper?
Model prices have historically fallen over time as newer models arrive and older ones are repriced, but nothing about Astra's future pricing has been announced, so treat that as a pattern rather than a forecast. Budget against the price today.
The short version
Not free, not on Free or Go, and the plan you need depends on whether you want it in Chat or in Work and Codex. On a subscription you pay monthly and live within a message cap; via the API you pay $10 per million input and $50 per million output tokens with no cap and no floor.
Do the arithmetic on your own volume before choosing, and check OpenAI's own pricing page for current availability — this rollout moved fast enough that any article's table is a snapshot. If you are going the API route, GPT-6 Astra API Cost Explained breaks the token math down properly, and GPT-6 Astra Explained covers whether you need the flagship at all.
Sources: OpenAI API model reference for gpt-6-astra and the GPT-5.6 family (per-token pricing); OpenAI Help Center and pricing pages for ChatGPT Go, Plus and Business plan prices; OpenAI GPT-6 Astra announcement (phased rollout across Chat, Work and Codex). Pro tier prices and per-plan weekly message caps are as reported at launch and should be confirmed in your own account.
