hi | OpenAI | Page 1

pulsar wagon Nov 14, 2025, 5:07 AM

#

hi

bleak flowerBOT Nov 14, 2025, 5:07 AM

#

Hi @pulsar wagon. I'm kapa.ai, your friendly support bot. I will try to answer your question by searching through the knowledge base. :wave:

#

Hello! How can I assist you with OpenAI today? If you have any questions about OpenAI's APIs, SDKs, or platform features, feel free to ask.

#

@pulsar wagon Kapa.ai is still learning and improving, please let me know how I did by reacting below.

pulsar wagon Nov 14, 2025, 5:07 AM

#

What are your limitations?

#

@bleak flower what are your limitations?

bleak flowerBOT Nov 14, 2025, 5:08 AM

#

Let me check the knowledge base for you again. One moment! :books:

#

OpenAI models, including ChatGPT, have several important limitations:

Knowledge Cutoff: The models are trained on data up to a certain point and do not know about events or developments after that date unless tools (like web search) are used. This means they may not provide up-to-date information on recent events or changes in knowledge domains.
Confidence vs. Accuracy: The model may express high confidence even when its answers are incorrect. Confidence in responses does not guarantee reliability.
Access Limitations: The model may not be able to access certain information due to technical issues, paywalls, or website restrictions (such as those set by robots.txt).
Bias and Oversimplification: Sometimes, the model may present a single perspective as absolute truth, oversimplify complex issues, or misrepresent the weight of scientific consensus or social debate.

#

Usage Limits: There are message and usage limits depending on your plan and the specific model you use. For example, o3 and o4-mini models have per-user message caps, and exceeding these limits will temporarily restrict access to those models until the quota resets. There are no overage fees, but you simply cannot use the model again until the limit resets. These limits are enforced per individual user, not at the team or workspace level for most models. There is currently no dashboard to track your usage, but you can see your reset date by hovering over the model name in the picker. Notifications are shown in-product when you hit a limit, and the model becomes unavailable until the reset time. Admins do not need to manage these limits for individual users. There are also API rate limits, which are set at the organization and project level, not the user level, and vary by model and usage tier. These limits help ensure fair access and system stability for all users. Violating usage policies, such as programmatically extracting data or sharing account credentials, is prohibited and may result in access restrictions or other actions by OpenAI. There are no penalties or extra fees for exceeding usage limits—just temporary unavailability until the quota resets. For more details, see the relevant help articles and documentation (ChatGPT Usage Limits, OpenAI o3 and o4-mini Usage Limits, API Rate Limits).

#

Feature Limitations: Some advanced features (like image generation, file uploads, or certain tools) may not be available in all models or plans, and some features have their own stricter rate limits, especially for free users.
Technical Limitations: Models can make mistakes, especially in less common environments or tasks. For example, the computer-use-preview model is less reliable for automating tasks on an OS compared to browser-based tasks.
No Fine-Tuning (for some models): Fine-tuning is not yet available for certain models like o3 and o4-mini.
Compliance and Data Handling: Some models are eligible for zero data retention (ZDR) for trusted customers, but not all features or models may support this.
Prohibited Uses: You may not use OpenAI services for illegal, harmful, or abusive activities, reverse engineering, programmatic data extraction, or to develop competing AI models. Sharing or reselling access is also prohibited.