Your privacy, your call.No advertising cookies.May we count your visit in aggregate (Cloudflare,no cookies) and remember where you came from?If you reject,we won’t measure your visit. · Cookies · Privacy
GPT-6 Astra vs Gemini 4 Argon: prices, API and privacy (2026)
GPT-6 Astra or Gemini 4 Argon: which plans include them, API rates (Argon’s price rises on 1 December), context, privacy and what data is missing. With official sources.
By Cognenza editorial team · Updated · Euro prices as published for Spain; they may differ in your country.
Verdict
Both are their makers’ flagship models and neither is on the free plan. If you pay for a subscription, choose by the ecosystem you already use: ChatGPT Plus (€23 per month) for Astra or Google AI Pro (€21.99 per month) for Argon. Via the API, Gemini 4 Argon is clearly cheaper at the official rates. We have not tested either of them ourselves, so we do not say which one answers better.
Which to choose for your profile
Students
Neither is on the free plan. You do not need a model in this range to study: start with the free tier of each service and only pay if it falls short.
Professionals
If you work in Gmail and Docs, Argon with Google AI Pro (€21.99 per month). If you work in ChatGPT and want documents, spreadsheets and presentations that respect your templates, Astra with Plus (€23 per month). OpenAI and Google do not publish usage limits for these models on those plans.
Freelancers and SMEs
For business: via the API, GPT-6 Astra costs USD 10.00 per million input tokens and USD 50.00 per million output tokens; Gemini 4 Argon costs USD 2.00 and USD 10.00 until 30 November, and USD 4.00 and USD 20.00 from 1 December. At those official rates, the same volume of tokens (input or output) costs 5 times more on Astra than on Argon until 30/11 and 2.5 times more from 1/12. Neither trains on your data in its paid API, and OpenAI lets you process Astra in Europe (with a 10% surcharge on the rate).
Developers
Similar context: 1,050,000 tokens on Astra (max. 128,000 output) and 1,048,576 on Argon (65,536 output by default, up to the full context with “long decode”). Argon accepts text, image, video, audio and PDF; Astra, text and image. Both return text only. Astra is also on Azure and Bedrock, and on Google Cloud in private access.
By budget
€0: Neither. Astra is not on ChatGPT Free or Go; Argon is not on free Gemini, Google AI Plus or the free API tier.
entry-level subscription: Astra from ChatGPT Plus (€23 per month); Argon from Google AI Pro (€21.99 per month).
heavy use: ChatGPT Pro (from €103 per month) adds Astra Pro; Google AI Ultra from €99.99 per month.
pay as you go (API): Argon, cheaper per token at the official rates (see “For business”); Argon’s rate doubles on 1 December 2026.
Main differences
Launch: GPT-6 Astra, 11 September 2026 (staggered rollout); Gemini 4 Argon, 13 October 2026.
App: Astra in ChatGPT Plus, Pro, Business and Enterprise; Argon in the Gemini app and AI Mode in Search with Google AI Pro and Ultra.
Input: Argon accepts text, image, video, audio and PDF; Astra, text and image. Both answer with text only via the API.
Privacy in the app: both may train on your chats by default and let you turn it off (ChatGPT: “Improve the model for everyone”; Gemini: activity saving). In the paid API, neither trains on your data.
Safety: OpenAI rates Astra at its highest cybersecurity capability level and adds monitoring that can pause tasks; Google applies filters, internal probes and a monitor that can stop the agent to Argon.
Side by side
GPT-6 Astra vs Gemini 4 Argon: prices, API and privacy (2026)
USD 2.00 per million input tokens (from 1 December 2026: USD 4.00)
Trains on your data?
Yes on consumer plans unless you turn it off; Business, Enterprise and Edu do not train by default
Yes, with activity saving turned on (with human reviewers; conversations they read are kept for up to 3 years)
How to avoid it
Settings > Data controls > “Improve the model for everyone” (also works without logging in; saved per browser). Temporary chats are not used for training and are kept for up to 30 days. Thumbs-up/down feedback may be used for training
Turn off activity saving: chats are kept for 72 hours and not used for training, unless you send feedback. Temporary chats: no training. Auto-delete by default: 18 months
Processing in Europe (EEA + Switzerland) available via the API, with a 10% surcharge
No verified data
Price on Google Cloud / Vertex AI
No verified data
No verified data
Common benchmark
No verified data
No verified data
Cognenza’s own test
No verified data
No verified data
API rates in US dollars (USD) per million tokens, as published by each company. Benchmarks are the manufacturer’s own figures: they are not Cognenza tests and cannot be compared with each other.
Benchmarks: manufacturer’s figures
Each maker publishes its own figures on different tests (OpenAI: FrontierMath, ARC-AGI-3, GPQA Diamond, OSWorld; Google: Vals Index, ERQA, Gray Swan). There is no common test published by both with the same methodology, so we do not compare them.
Manufacturer’s figure, not comparable across models.
Our own tests
We have not tested either of them ourselves. The only no-sign-up tests we ran were on free ChatGPT without logging in, where you cannot see which model answers; those tests are not of GPT-6 Astra. Testing Astra or Argon requires a subscription or a paid API, and for now we only run free, no-sign-up tests.
Unverified data
usage limits on Plus/Pro and Google AI Pro/Ultra: No verified data (not published by OpenAI or Google)
Gemini EU data residency (app and API): No verified data (not published in the sources we checked)
Astra’s price on Google Cloud: No verified data (not published (private access))
common benchmark: No verified data (no test published by both with the same methodology)
our own test: No verified data (requires payment or sign-up; current rule: free and no sign-up only)
Gemini 4 Argon price on Vertex AI: No verified data (Google points to another pricing page, not checked)
Frequently asked questions
Can I use GPT-6 Astra or Gemini 4 Argon for free?
No. Astra starts at ChatGPT Plus and Argon at Google AI Pro; Argon is not on the free API tier either.
Which is cheaper via the API?
Gemini 4 Argon: USD 2.00 input and USD 10.00 output per million tokens until 30/11 (then USD 4.00 and USD 20.00), compared with USD 10.00 and USD 50.00 for GPT-6 Astra.
Which is better?
We do not know from our own data: we have not tested them, and each maker’s figures use different tests.
Where are they available?
Google launched Argon globally and OpenAI offers Astra on its paid plans; via the API, OpenAI lets you process Astra in Europe.
Can I use GPT-6 Astra on Google Cloud?
Since 21/10/2026 it has been in private access for some companies; general availability is expected in the coming weeks and there is no published price yet.
Model by model
GPT-6 Astra at a glance
Release
11 September 2026
Knowledge cut-off
30 April 2026
Context
1,050,000 tokens
Input
text, image
Output
text
Consumer plans
Plus: €23 per month
Pro: from €103 per month
Where it is available
OpenAI API, Microsoft Azure, Amazon Bedrock; Google Cloud (Model Garden) in private access for selected customers since 21/10/2026, general availability “in the coming weeks”, no published price
API privacy
The API does not train on your data unless you agree to share it. Astra supports Zero Data Retention for eligible customers. gpt-6-astra is listed among the endpoints with processing in Europe (EEA + Switzerland); Fast mode is not available with EU residency
Safety
Meets the “Critical” cybersecurity threshold of OpenAI’s Preparedness Framework; monitoring that can pause or stop tasks
Yes on consumer plans unless you turn it off; Business, Enterprise and Edu do not train by default
How to avoid it
Settings > Data controls > “Improve the model for everyone” (also works without logging in; saved per browser). Temporary chats are not used for training and are kept for up to 30 days. Thumbs-up/down feedback may be used for training
Yes, with activity saving turned on (with human reviewers; conversations they read are kept for up to 3 years)
How to avoid it
Turn off activity saving: chats are kept for 72 hours and not used for training, unless you send feedback. Temporary chats: no training. Auto-delete by default: 18 months
API rates (USD per million tokens): Gemini 4 Argon
Model
Input
Cached input
Output
gemini-4-argon (Launch price until 30/11/2026. Output includes reasoning tokens. No free tier. Batch and Flex: half price. Google Search: 5,000 free queries a month, then $14 per 1,000)
Every figure comes from the company’s official page and shows the date we checked it. If something is not officially published, we say so; we never estimate it. How we work.
Weekly AI newsletter
Every Tuesday at 07:00 UK time: the week’s AI news with sources, a guide and a tool. Free, no ads, no ad tracking. Unsubscribe in one click. What is inside and a sample issue
Take Cognenza with you
Install it as an app: opens instantly, works offline with what you have already read and asks for no data.
