ChatGPT Business and Enterprise Limits Explained
Managed ChatGPT plans combine broad base-model access with model-specific quotas, workspace controls, and flexible credits. Organizations therefore need to manage usage by model and feature rather than by total conversations alone.
Quick Answer: ChatGPT Business and Enterprise provide virtually unlimited eligible base-model messages, but reasoning models, Pro models, agents, research, images, voice, and office extensions can have separate allowances or credit costs. Workspace settings and the current rate card are the source of truth.
What does virtually unlimited mean?
OpenAI says Business and Enterprise offer virtually unlimited messages for eligible base models. This access is subject to abuse guardrails and the Services Agreement.
It does not mean all advanced models and features are included without limits. Reasoning, Pro access, agentic work, research, image generation, voice, and office-extension tasks can use separate counters or credits.
What Business limits are currently published?
OpenAI’s Business limits article currently lists GPT-5.5 Instant as virtually unlimited, GPT-5.5 Thinking at 3,000 requests per week, and GPT-5.5 Pro at 15 requests per month.
The page also says model availability and limits change over time. The workspace model picker and current flexible-pricing rate card should be checked during model transitions.
What are the current workspace context windows?
OpenAI currently documents 128K context for GPT-5.6 Luna and Terra and 272K for GPT-5.6 Sol in ChatGPT Business.
Model-picker labels such as Medium, High, and Extra High change reasoning effort but do not by themselves change context size, allowance, or credit rate.
How does flexible credit pricing work?
The current rate card lists GPT-5.6 Sol at about 10 credits per message and GPT-5.6 Sol Pro at 50 credits. Instant can automatically route to Medium, which then consumes 10 credits.
Feature rates currently include 30 credits per Agent mode message, 50 per Deep Research task, 5 per image generation, and 5 per voice minute.
How are office and agent tasks charged?
Excel or Sheets, PowerPoint, and Workspace Agents use token-based pricing rather than one fixed credit amount. OpenAI estimates a typical Excel or Sheets task at 5–20 credits, a PowerPoint task at 10–50, and a GPT-5.5 Workspace Agent run at 5–25.
ChatGPT for PowerPoint is documented as free for Business and Enterprise through August 6, 2026, after which token-based charging begins.
How is Enterprise access controlled?
Enterprise and Edu model access is governed by workspace settings and RBAC. Admins can control reasoning access, Pro access, and automatic-routing options.
A model rollout does not automatically override the organization’s settings. Two Enterprise workspaces may therefore show different options and effective limits.
What should administrators measure?
Administrators should separate base messages, reasoning requests, Pro requests, agent tasks, research, images, voice, office extensions, and failed or repeated workflows.
The most useful metric is cost per verified business outcome, not raw message volume. This is especially important as agentic features consume credits based on task complexity.
Related PiSkill Resources
Explore the Subscription Spend Optimization Analyst, Compliance Evidence Collection Automation Designer, LLM Evaluation Benchmark and Regression Designer, and Tool Selection and Agent Routing Architect skills.
Sources
- OpenAI: Business Models and Limits
- OpenAI: Enterprise and Edu Models and Limits
- OpenAI: ChatGPT Rate Card