Workshops Use Cases Architecture Projects Blog
About Contact Offers Talks
Home›AI tools›ChatGPTDeutsch
openaicodex

ChatGPT: one vendor, OpenAI models only, several accounts that can serve them

ChatGPT is OpenAI's assistant, with Chat for conversations and Work for longer tasks. Codex is its coding agent, and the OpenAI API serves the same models to developers.

Book a free 30-min call →AI Enablement Workshop →

OpenAI sorts its own product by the kind of work: Chat for questions and drafts, Work for multi-step tasks that end in a finished deliverable, and Codex for software. Since July 2026 all three sit in one desktop app, and you switch between them in the menu.

Only OpenAI models run here. OpenAI decides which model family a plan gets; you choose a thinking level in Chat or a model in Work and Codex, and Business and Enterprise admins can limit the choice. ChatGPT itself always runs on OpenAI's own service. The models behind it can also be served by your own account: the OpenAI API, Microsoft Foundry, Amazon Bedrock, or, for the open-weight gpt-oss models, your own hardware.

The family

ChatGPT (Chat)[1]

Web, iOS, Android, macOS, Windows
The general assistant for questions, files, images, voice, Projects and memory.

ChatGPT Work[2]

Web, mobile, desktop app
An agent "designed for longer, multi-step work and finished deliverables", which can use connected apps and run scheduled tasks.

ChatGPT desktop app[3]

macOS, Windows
One app with Chat, Work and Codex; the previous app stays available as "ChatGPT Classic".

Codex[4]

Desktop app, web, cloud, GitHub, Slack, SDK
OpenAI's coding agent: it writes and debugs code, runs tests and commands, and reviews changes.

Codex CLI and IDE extension[5]

CLI, IDE
Codex in the terminal and the editor; the CLI is open source under Apache-2.0.

ChatGPT Atlas[6]

Browser, macOS (Apple silicon)
OpenAI's own web browser with ChatGPT built in.

OpenAI API[7]

API
Pay-per-use access to the same models for your own applications, including the Agents API in public beta.

gpt-oss[8]

Self-hosted, partner clouds
Open-weight models (gpt-oss-120b, gpt-oss-20b) under Apache 2.0 that you run yourself; they are not part of ChatGPT or the OpenAI API.

Plans

As of September 2026, ChatGPT has four personal plans and three plans for organisations; the API is billed separately by use.

PlanForWhat changes
Free[9]Individuals trying ChatGPT.GPT-5.6 Luna; limited uploads, images, voice, memory and Codex. Content can be used for training unless you opt out. No admin features.
Go[9]Individuals who need higher limits than Free.Same model as Free with more usage and longer memory. Consumer data terms, no admin features.
Plus[10]Individuals using ChatGPT for daily work.GPT-5.6 Sol in Chat, GPT-6 Astra inside Work and Codex, apps and Projects. Consumer data terms, no admin features.
Pro[11]Individuals with heavy research or coding use.Two tiers with the same features and 5x or 20x the usage of Plus, including GPT-6 Pro. Consumer data terms, no admin features.
Business[12]Teams, from two seats (formerly called Team).Shared workspace, admin console and SAML SSO; workspace data is not used for training. Standard and Premium seats can be mixed.
Enterprise[13]Larger organisations, through OpenAI sales.Adds SCIM, role-based access, compliance logs, IP allowlisting, customer-managed keys, retention control and data residency, including Europe.
Edu[14]Universities.Enterprise-style admin and data terms for campuses, including data residency for new workspaces.

Which model runs

OpenAI decides which model family each plan and surface gets. In Chat you pick a thinking level, in Work and Codex you pick a model, and in Business and Enterprise workspaces admins can restrict which models members use.

GPT-6 Astra[10]Most capable; shown as "GPT-6 Pro" in Chat on Pro, Business and Enterprise; in Work and Codex from Plus up
GPT-6 Sol[15]Work, Codex and API, since 22 Sep 2026
GPT-6 Luna[15]Efficient model for Work, Codex and API, since 22 Sep 2026
GPT-5.6 Sol and Sol Pro[10]Chat on paid plans, with a thinking slider
GPT-5.6 Terra[10]Balance of capability, speed and cost
GPT-5.6 Luna[10]Default model on Free and Go
Codex model choice[16]Recommended model by default; change it with --model or config.toml, or point Codex at another compatible provider
gpt-oss-120b, gpt-oss-20b[8]Open weights, run on your own hardware or a partner cloud

What you can configure

SSO and SCIM[13]SAML SSO on Business, Enterprise, Edu and API. SCIM provisioning only on Enterprise and Edu.
Roles and groups (RBAC)[17]Enterprise and Edu grant features, apps and developer mode per role. Business has workspace-wide switches only.
Apps (connectors) and plugins[17]Apps connect ChatGPT to services such as Google Drive or Slack. They are on by default in Business and off by default in Enterprise and Edu, with separate switches for read and write actions.
Custom MCP apps[18]Developer mode with full MCP support, including write actions, is in beta for Business, Enterprise and Edu; admins decide who may use it.
Model availability[19]Admins in Business and Enterprise can limit which models members use, and Enterprise sets default model and reasoning level for Work and Codex.
Memory[20]Saved memories and reference to past chats. Users turn it off in settings; workspace admins manage memory settings per role.
Projects[21]Projects keep related chats, files and instructions together, on every plan. Shared projects in Business and Enterprise only use memory from inside the project.
Codex access[22]Enterprise admins decide which users or groups get Codex and at what level. Codex on an API key has no cloud features such as GitHub code review or Slack.
Compliance logs[23]Enterprise exports conversations, files, memories, users, admin audit and Codex usage as log files for tools such as Microsoft Purview.
Retention and encryption[13]Qualifying organisations set how long OpenAI keeps business data. Enterprise Key Management lets customers hold their own encryption keys.

What happens to your data

Personal plans (Free, Go, Plus, Pro)[24]OpenAI states "we may use your content to train our models". You can opt out in Data controls or the Privacy Portal. Rating an answer can still share that conversation.
Deleting chats[25]Deleted chats are removed within 30 days, unless OpenAI must keep them longer for security or legal reasons. Temporary chats are kept up to 30 days.
Business, Enterprise, Edu[13]Inputs and outputs are not used for training by default. Enterprise and Edu admins control retention.
OpenAI API[26]API data is not used for training unless you opt in. Abuse monitoring logs are kept up to 30 days; zero data retention needs OpenAI approval.

Europe: GDPR, data residency and the AI Act

DPA and GDPR terms[27]OpenAI signs a Data Processing Addendum with Standard Contractual Clauses for Business, Enterprise and the API. OpenAI Ireland Limited processes data from the EEA and Switzerland.
Personal plans in the EU[28]OpenAI Ireland Limited is the controller. OpenAI names legitimate interests as the basis for training, and users can opt out.
Data residency in Europe[14]Since February 2025, new Enterprise and Edu workspaces and approved API projects can store data at rest in Europe (EEA and Switzerland). Processing in Europe is also offered. Codex Web, third-party GPTs and beta features are excluded. Not available on personal plans or Business.
AI Act[29]OpenAI is listed by the European Commission as a signatory of the General-Purpose AI Code of Practice.

What the AI Act and the GDPR ask of you as a user →

Hybrid: the same models in your own account

ChatGPT itself runs only on OpenAI's service. The models behind it can run under your own cloud account, region and identity, with limits that differ by route.

OpenAI API with EU data residency[26]An API project with European data residency keeps data at rest in Europe, and since August 2026 single requests can be processed in Europe. This needs approval from OpenAI.
Microsoft Foundry (Azure OpenAI)[30]GPT-6 and GPT-5.6 models run in your Azure tenant as models sold by Azure. Microsoft states that prompts and outputs "are NOT available to OpenAI".
Foundry and the EU[31]Microsoft's blog of 22 Sep 2026 says GPT-6 standard deployments cover US and EU Data Zones. The Microsoft Learn region table does not show the EU yet (checked 25 Sep 2026).
Amazon Bedrock[32]GPT-6 Astra (us-west-2), GPT-6 Sol and GPT-6 Luna (us-east-1) run on AWS infrastructure, and AWS does not share content with OpenAI in the default modes. No EU region is documented, and OpenAI notes that an AWS Region is not an OpenAI data residency jurisdiction.
gpt-oss on your own hardware[8]gpt-oss-120b fits a single 80 GB GPU, gpt-oss-20b needs 16 GB of memory. OpenAI does not receive the data you send to self-hosted models.
gpt-oss in the cloud[33]Both gpt-oss models are offered on Amazon Bedrock, Google Cloud Vertex AI and Microsoft Foundry.
Codex with your own key or provider[22]Codex CLI, SDK and IDE extension can use an API key or any provider with a Chat Completions or Responses API; cloud features are then not available.
The same finished product, in three configurations: consumer, business, or on a cloud account you already have.
The same finished product, in three configurations: consumer, business, or on a cloud account you already have.[34]
Recently changed
  • ChatGPT agent (agent mode) is no longer available; ChatGPT Work took over multi-step tasks, launched 9 Jul 2026.
  • Custom GPTs are being replaced by plugins; Enterprise workspaces lose them on 11 Dec 2026.
  • GPT-6 Sol and GPT-6 Luna arrived in Work, Codex and the API on 22 Sep 2026.
Linda Mohamed
How I use it
I use ChatGPT to generate images.

What would help you next?

Learn to decide

The course “From Managed to Hybrid” shows where AI should run: what each layer costs, what you operate and where your data has to stay. In preparation, with a free live session first.

Join the course waitlist →
Linda MohamedLinda Mohamed
AI and cloud consultant, AWS Community Hero, Vienna

Roll it out with me

Tell me which tools your teams use or want. In 30 minutes we look at plans, settings, data rules and whether AWS funding can cover part of a pilot. How AWS funding works →

Book a free 30-min call →15-min question slot

You already know what you need

Book a workshop directly:

  • AI Enablement Workshop →
    One workshop for one role, up to 12 people: how your team uses these tools well, and what stays inside the rules.
  • AI Discovery Workshop →
    Sort your ideas into a scored shortlist, so you know which use case is worth more than a tool licence.
ai-solutions.wiki

Read more on ai-solutions.wiki

My open knowledge base, with longer comparisons and sources.

The technology underneath: AI technologies explained →

Sources

  1. ChatGPT (Chat), chatgpt.com. https://chatgpt.com/download/
  2. ChatGPT Work, help.openai.com. https://help.openai.com/en/articles/20001275-chatgpt-work-and-codex
  3. ChatGPT desktop app, help.openai.com. https://help.openai.com/en/articles/20001276-moving-to-the-new-chatgpt-desktop-app
  4. Codex, learn.chatgpt.com. https://learn.chatgpt.com/docs
  5. Codex CLI and IDE extension, github.com. https://github.com/openai/codex
  6. ChatGPT Atlas, help.openai.com. https://help.openai.com/en/articles/12603091-chatgpt-atlas-for-enterprise
  7. OpenAI API, developers.openai.com. https://developers.openai.com/api/docs/changelog
  8. gpt-oss, help.openai.com. https://help.openai.com/en/articles/11870455-openai-open-weight-models-gpt-oss
  9. Free, chatgpt.com. https://chatgpt.com/pricing/
  10. Plus, help.openai.com. https://help.openai.com/en/articles/20001354-gpt-56-in-chatgpt
  11. Pro, help.openai.com. https://help.openai.com/en/articles/9793128-about-chatgpt-pro-tiers
  12. Business, help.openai.com. https://help.openai.com/en/articles/8792828-chatgpt-business-overview
  13. Enterprise, openai.com. https://openai.com/business-data/
  14. Edu, help.openai.com. https://help.openai.com/en/articles/9903489-data-residency-and-inference-residency-for-chatgpt
  15. GPT-6 Sol, openai.com. https://openai.com/index/introducing-gpt-6-sol-and-luna/
  16. Codex model choice, learn.chatgpt.com. https://learn.chatgpt.com/docs/models
  17. Roles and groups (RBAC), help.openai.com. https://help.openai.com/en/articles/11509118-admin-controls-security-and-compliance-in-apps-enterprise-edu-and-business
  18. Custom MCP apps, help.openai.com. https://help.openai.com/en/articles/12584461-developer-mode-and-mcp-apps-in-chatgpt
  19. Model availability, help.openai.com. https://help.openai.com/en/articles/20001264-chatgpt-enterprise-admin-quickstart
  20. Memory, help.openai.com. https://help.openai.com/en/articles/8590148-memory-faq
  21. Projects, help.openai.com. https://help.openai.com/en/articles/10169521-projects-in-chatgpt
  22. Codex access, learn.chatgpt.com. https://learn.chatgpt.com/docs/pricing
  23. Compliance logs, openai.com. https://openai.com/index/new-tools-for-chatgpt-enterprise/
  24. Personal plans (Free, Go, Plus, Pro), help.openai.com. https://help.openai.com/en/articles/5722486-how-your-data-is-used-to-improve-model-performance
  25. Deleting chats, help.openai.com. https://help.openai.com/en/articles/8809935-how-to-delete-and-archive-chats-in-chatgpt
  26. OpenAI API, developers.openai.com. https://developers.openai.com/api/docs/guides/your-data
  27. DPA and GDPR terms, openai.com. https://openai.com/policies/data-processing-addendum/
  28. Personal plans in the EU, openai.com. https://openai.com/policies/eu-privacy-policy/
  29. AI Act, digital-strategy.ec.europa.eu. https://digital-strategy.ec.europa.eu/en/policies/contents-code-gpai
  30. Microsoft Foundry (Azure OpenAI), learn.microsoft.com. https://learn.microsoft.com/en-us/azure/foundry/responsible-ai/openai/data-privacy
  31. Foundry and the EU, azure.microsoft.com. https://azure.microsoft.com/en-us/blog/gpt-6-astra-sol-and-luna-for-production-agents-in-microsoft-foundry/
  32. Amazon Bedrock, developers.openai.com. https://developers.openai.com/api/docs/guides/amazon-bedrock
  33. gpt-oss in the cloud, docs.aws.amazon.com. https://docs.aws.amazon.com/bedrock/latest/userguide/model-parameters-openai.html
  34. Linda Mohamed, course slides “Start here”, MASE-3, FH Burgenland, September 2026
  35. What is ChatGPT, ai-solutions.wiki. https://ai-solutions.wiki/basics/what-is-chatgpt/
  36. ChatGPT Free vs Plus vs Pro, ai-solutions.wiki. https://ai-solutions.wiki/comparisons/chatgpt-free-vs-plus-vs-pro/
  37. Claude Code vs Cursor vs Codex, ai-solutions.wiki. https://ai-solutions.wiki/comparisons/claude-code-vs-cursor-vs-codex/
  38. ChatGPT vs Microsoft Copilot, ai-solutions.wiki. https://ai-solutions.wiki/comparisons/chatgpt-vs-microsoft-copilot/
  39. Microsoft Learn: Foundry region availability (EU Data Zone not shown for GPT-6, checked 25 Sep 2026). https://learn.microsoft.com/en-us/azure/foundry/foundry-models/concepts/models-sold-directly-by-azure-region-availability
  40. Google Cloud: OpenAI gpt-oss models. https://docs.cloud.google.com/vertex-ai/generative-ai/docs/maas/openai
  41. Microsoft Learn: Foundry Models sold by Azure (gpt-oss). https://learn.microsoft.com/en-us/azure/foundry/foundry-models/concepts/models-sold-directly-by-azure
  42. ChatGPT agent is no longer available. https://help.openai.com/en/articles/11752874-chatgpt-agent
  43. ChatGPT Work launch, 9 Jul 2026. https://openai.com/index/chatgpt-for-your-most-ambitious-work/
  44. Custom GPT retirement dates. https://help.openai.com/en/articles/10128477-chatgpt-enterprise-and-edu-release-notes
  45. Premium seats in ChatGPT Business. https://openai.com/index/premium-seats-chatgpt-business/
  46. Apps and plugins in ChatGPT. https://help.openai.com/en/articles/11487775-connectors-in-chatgpt
  47. OpenAI API models. https://developers.openai.com/api/docs/models
  48. Introducing data residency in Europe. https://openai.com/index/introducing-data-residency-in-europe/
  49. OpenAI on the EU codes of practice. https://openai.com/index/supporting-eu-trustworthy-ai-ecosystem/

Checked against the vendors' own pages in September 2026. Plans, models and features change often: check the linked pages before you decide.