Body
CES is introducing a standardized OpenAI service to provide broad, responsible access to AI across the organization. The service is designed to support real institutional work while evolving with CES needs, policies, and experience.
For most users, the transition should be straightforward. You do not need to understand credits, tokens, or the technical details of AI models to begin using ChatGPT.
Understanding the new service
What is changing?
CES is moving from a project-based rollout to a more consistent CES OpenAI service. The goal is to make AI broadly available while providing a secure, supported environment for appropriate institutional use. For most users, no action is required before the service reaches their organization or campus. Onboarding communication will explain when access is available and any steps that apply.
What features are included in the new service?
The CES Education core includes GPT-5.5 Instant; file uploads and analysis; Canvas and code execution; web search; projects; custom GPT creation; Library + Memory; and data export. Additional capabilities may be introduced as the service evolves, including image generation, reasoning models, Deep Research, voice, recording and transcript reference, Office add-ins, connected data, Skills, Agents and Sites, Work in ChatGPT, and Codex developer workflows.
Feature availability reflects CES role policy and may change as capabilities and institutional practices evolve.
Will everyone have exactly the same access?
Not necessarily. Some capabilities may be enabled differently as CES aligns access with institutional roles, policies, security controls, and demonstrated work needs. Differences in access are not a judgment about a person's position, importance, or ability to use AI.
Does the new service limit what I can use ChatGPT for?
The service supports appropriate institutional use of ChatGPT, subject to CES policies and guidance. Some activities and features may have availability or usage limits. These limits help CES provide broad access while managing the service responsibly.
Usage, credits, and capacity
What are credits?
Credits are a way of managing AI usage across the CES OpenAI environment. Different AI activities may consume different amounts of capacity. More computationally intensive models or activities can consume more than simpler interactions. For most users, the important point is: You do not need to calculate credits before using ChatGPT. Use the service normally. The credit system helps CES understand and manage overall consumption.
Are credits the same as tokens?
No. Tokens are a technical measurement used by AI models to process text and other information. Credits are the higher-level mechanism being used to manage available AI usage. Most users do not need to manage tokens directly.
What happens if I reach a usage limit?
If a usage limit prevents you from completing your work, CES has a process for evaluating additional access. Don't avoid productive uses of AI because you're worried about credits. Normal usage also helps CES see where more capacity is needed.
If you've already been approved for more credits and used up your premium credits, the CES contract with OpenAI allows unlimited use of these models:
| Product |
Unlimited model |
| ChatGPT |
Instant |
| ChatGPT Work |
Luna |
| Codex |
Luna |
Here, Instant and Luna are model tiers.
Can I request additional capabilities or capacity?
Yes. If your work requires capabilities or capacity beyond what is currently available to you, an established request or exception process can be used to evaluate that need. Additional information about that process will be communicated as part of the rollout.
Should I request additional access immediately?
Generally, no. Start by using the new service for appropriate work. This will help both you and CES understand actual usage requirements. If you encounter a meaningful limitation that prevents you from completing your work, that is a better indication that additional access may be appropriate.
Will everyone with the same job title have identical access?
Not necessarily. AI requirements can vary significantly even among people with similar roles. Actual work requirements and usage may differ. The service model is designed to accommodate those differences over time.
Will the service stay the same permanently?
Not necessarily. CES expects the service model to evolve as we learn more about how faculty, staff, students, and other users employ AI. Included capabilities, usage allocations, and access processes may therefore change over time.
Which model should I use?
Frontier AI models are genuinely expensive to run, and everyone at CES draws from the same shared pool of capacity. Choosing the right model for each task, instead of always reaching for the most powerful one, is what keeps broad access available to everyone.
A good rule of thumb: pick the least powerful model that still does the job well. Match the model to the difficulty of the task, not to "always use the strongest." Reaching for the most capable model by default burns through shared capacity faster than most work requires.
If you're not sure where to start, use Terra (GPT-5.6-Terra). It's the recommended default for most everyday work.
The three models
| Model |
Best for |
| Terra (GPT-5.6-Terra) |
Your everyday default. Balances intelligence and cost, and handles most tasks well. |
| Luna (GPT-5.6-Luna) |
Cost-sensitive, high-volume work. The most economical model, and a good way to stretch your credits. |
| Sol (GPT-5.6-Sol) |
The most demanding, complex tasks. The most powerful model, and the one that uses credits fastest, so save it for work that truly needs it. |
When to use which
These are illustrative examples to give you a feel for the right fit, not an official or complete list:
- Terra: Drafting an email, summarizing a document, brainstorming ideas, or answering quick questions.
- Luna: Large or repetitive jobs, high-volume work, or times when you want to conserve credits.
- Sol: Genuinely hard, multi-step reasoning or complex analysis that the other models can't handle well.
| Tip: Running low on credits? Switch to Luna to keep working, since it's the most economical model. Reserve Sol for tasks that truly need it, because it consumes credits the fastest. |
All of the latest OpenAI models understand text and image input, respond in text, and work in multiple languages.
Accounts, appropriate use, and training
Do I need to create a new ChatGPT account?
Follow the onboarding instructions provided for your CES OpenAI environment. It is important to distinguish between your institutional ChatGPT workspace and any personal ChatGPT account you may already have. Institutional work should be performed using the CES-provided environment when required by CES policy and guidance.
What happens to my personal ChatGPT account?
Your personal ChatGPT account and your CES-provided access should be treated as separate environments unless specific migration instructions are provided. Having CES-provided access does not necessarily mean that your personal account, conversations, settings, or subscription automatically become part of the CES environment.
What should I use ChatGPT for?
ChatGPT can assist with many common activities, including:
- Brainstorming and exploring ideas
- Drafting and revising content
- Summarizing information
- Research and analysis
- Preparing presentations and communications
- Working with documents and data
- Coding and technical work
- Learning new concepts
- Improving routine workflows
The objective of the rollout is not simply to provide another technology tool. It is to help faculty and staff identify responsible, useful ways AI can improve their work.
What information can I put into ChatGPT?
Use the CES-provided environment according to institutional privacy, security, data-handling, and acceptable-use requirements. CES guidance should determine what institutional information may be used with AI. When uncertain about sensitive or restricted information, consult the applicable CES or campus guidance before submitting it.
Will training be available?
Yes. Training and awareness resources are part of the rollout. Users should expect guidance covering both how to use AI effectively and how to use it responsibly within CES. Additional resources will continue to be published as capabilities and institutional practices evolve.
Why is CES using a managed service instead of giving everyone unlimited access?
Enterprise AI resources have real capacity and cost implications, and usage can vary dramatically between users.
A managed service allows CES to:
- Provide AI access broadly
- Give most users enough capacity for normal work
- Provide additional resources where there is demonstrated need
- Understand how AI is actually being used
- Manage institutional costs responsibly
- Adjust the service as AI capabilities and usage patterns change
Getting started
What do I need to do right now?
For most users:
- Watch for your onboarding communication. It will explain when access is available and any steps you need to complete.
- Begin using the CES OpenAI service. You do not need to determine a service category before getting started.
- Complete the recommended or required AI training. Understand both the capabilities of the tools and CES expectations for responsible use.
- Use AI for real work. Experiment with appropriate tasks and identify where it can improve your effectiveness.
- Request additional capacity only when your work demonstrates a need for it.
The most important thing to understand
The CES OpenAI service is designed to help you do your work, not create another administrative hurdle. CES is expanding AI access so that people can learn how these tools can improve teaching, learning, administration, development, and other institutional work.
Start with the access provided to you, learn how to use it effectively, and allow your actual work requirements to determine whether additional AI capacity is needed.