ChatGPT Edu at BYU-Idaho uses a shared credit pool within the CES ChatGPT workspace. Advanced features are metered in two different ways, and knowing which one applies explains what users see on their Usage page.
| Important: A common misunderstanding is that every advanced feature draws from monthly credits from the first use. In the CES workspace, ChatGPT Work and Codex use an included 5-hour and weekly allowance first. Other advanced features use credits from the shared pool every time. |
The Short Version
- Work and Codex: count against an included 5-hour and weekly allowance first. Monthly credits do not change while that allowance remains.
- Other advanced features (Sol and Pro chat models, agent mode, deep research, image generation, and voice): use credits from the shared pool on every use, after any free daily allowance.
- Core features (Instant, web search, file uploads, canvas): do not use credits.
- Monthly usage limit: an admin-set cap on how many credits a user may use from the shared pool.
The Three Bars on the Usage Page
| Bar |
What it counts |
Who sets it |
| Monthly usage limit |
Credits used from the shared CES pool this month |
CES workspace admins, by tier, group, or user |
| 5 hour usage limit |
Work and Codex usage within a 5-hour block |
Not configurable in CES workspace admin settings |
| Weekly usage limit |
Work and Codex usage over a weekly period |
Not configurable in CES workspace admin settings |
How Work and Codex Are Metered
- ChatGPT Work and Codex share the same pricing, credits, and usage limits.
- Usage counts against the 5-hour and weekly allowance first. In a CES workspace test on October 16, 2026, several light Codex tasks on GPT-6.1 Sol lowered the 5-hour bar by 4% and the weekly bar by 1%, with no change to monthly credits used.
- The 5-hour block starts with the first Work or Codex task and resets 5 hours later. The Usage page shows the reset time.
- Fast mode uses the included allowance at 2.5× the Standard rate. Ultrafast (GPT-6 Astra and GPT-6.1 Sol) uses it at 8×.
- ChatGPT for Excel, PowerPoint, and Word, and Workspace Agents share agentic usage and credits with Work and Codex where available.
| Note: What happens when the 5-hour or weekly allowance runs out is not yet confirmed for the CES workspace. OpenAI's Codex pricing page says, "After you reach your included limits, available credits let you continue working." Until this is confirmed, check both the allowance bars and the monthly bar when a user reports that Work or Codex stopped. |
| Note: OpenAI's general documentation says Enterprise and Edu workspaces on flexible pricing have no fixed rate limits. The CES workspace does show and count 5-hour and weekly limits for Work and Codex, so do not apply that general statement to CES. |
How Other Advanced Features Are Metered
These features use credits from the shared pool on every use. Rates are approximate; see ChatGPT Edu - Credit Model for Codex and ChatGPT Work for the full rate card.
| Feature |
Credits |
| Sol chat message (Medium, High, or Extra High reasoning) |
10 per message |
| Pro chat message |
50 per message |
| Agent mode |
30 per message |
| Deep research |
50 per task; the first 5 per rolling 24 hours are free |
| Image generation |
5 per image; the first 3 per rolling 24 hours are free |
| Voice |
1.25 per minute |
Two routing rules affect chat costs. Instant may switch to Medium for requests that need more reasoning, and that message bills 10 credits. When ChatGPT routes a message to a mini model, it uses 0 credits.
Monthly Usage Limits
Monthly limits serve a different purpose.
Enterprise administrators may assign users or groups a monthly usage limit. This determines how much of the organization's shared credit pool they are authorized to consume during the usage period.
Think of it this way:
Organization's Shared Credit Pool
│
├── User A consumes credits
├── User B consumes credits
├── User C consumes credits
│
└── Individual or group limits control
how much each user may consume
A monthly limit does not provide additional credits. It is a guardrail on access to the existing shared pool.
A user can therefore reach their personal usage limit even while credits remain available to the organization as a whole. If the organization's shared pool runs out, advanced features pause for everyone unless workspace owners enable overages or purchase more credits.
What Does "Consumption" Mean?
Different ChatGPT features consume different amounts of resources.
For example, usage may vary based on:
- The model being used
- The length of the conversation or context
- If “Reasoning” is enabled
- Input and output token volume
- Tool use
- Work or Codex activity
- Agent usage
- Other computational requirements of the task
Frequently Asked Questions
Do Work and Codex use my monthly credits right away?
No. Work and Codex use the included 5-hour and weekly allowance first. Monthly credits do not change while that allowance remains.
What happens when my 5-hour or weekly allowance runs out?
This is not yet confirmed for the CES workspace. OpenAI says available credits let you continue working after the included limits are reached. The Usage page shows when each allowance resets.
Why are my monthly credits going down when I am not using Codex?
Sol and Pro chat messages, agent mode, deep research, image generation beyond the free daily allowance, and voice all use monthly credits on every use. Instant may also switch to Medium for harder requests, which bills 10 credits.
Can workspace admins change the 5-hour or weekly limits?
No setting for these limits is available in the CES workspace admin settings. Admins set monthly usage limits.
When does the 5-hour limit reset?
Five hours after the first Work or Codex task in the current block. The Usage page shows the exact reset time.
What does my monthly limit mean?
Your monthly limit is an administrative usage guardrail. It determines how much of the organization's shared credit pool you are permitted to consume during the applicable usage period.
It does not mean that the organization has reserved that exact number of credits specifically for you.
If I don't use my monthly allocation, do those credits belong to me?
No. The organization operates from a shared credit pool. Individual limits govern access to that pool; they do not create individual pools of credits.
Can another user's activity affect the organization's available credits?
Yes. Because credits are shared across the Enterprise environment, eligible activity from users across the organization contributes to total credit consumption. Individual and group usage limits help manage that shared resource.
Why can two similar prompts consume different amounts?
Credit consumption depends on the computational work required to complete the request. Factors can include model choice, input size, generated output, conversation context, reasoning effort, tool usage, and the specific feature being used.
Prompt count alone is therefore not a reliable measure of credit consumption.
Troubleshooting Tips
- Work or Codex stopped or slowed, but monthly credits remain: check the 5-hour and weekly bars and their reset times.
- Monthly credits are dropping quickly: check for Sol or Pro chat, agent mode, deep research, image generation, or voice use.
- User needs a higher monthly limit: see How to Request More ChatGPT Edu Usage.
Key Takeaway
| Important: Match the bar to the work. For Work and Codex, watch the 5-hour and weekly bars. For Sol or Pro chat, agent mode, deep research, images, and voice, watch the monthly credits. |
Further Reading