TinyMCE AI Limits

Overview

TinyMCE AI applies limits on API requests, context and attachments, models, and files to ensure fair usage, optimal performance, and cost control. The sections below describe rate limits, context limits, model-specific constraints, and file restrictions.

Rate Limits

Rate limits control the frequency of requests to prevent abuse and ensure service stability. These limits apply to both plugin usage and direct API calls, as the plugin uses the API under the hood. The service implements limits on requests, token usage, web search, and web scraping requests per minute. All rate limits are applied at both organization level (higher limits) and individual user level (lower limits) to ensure fair usage.

Specific rate limit values are subject to change and may vary based on the subscription tier. For current rate limit details for the environment, contact Tiny Support.

Context Limits

Context limits control how much content can be attached to conversations to ensure AI models can process all information effectively. These limits vary by model based on their specific capabilities and processing requirements.

File Limits

Supported attachment types include PDF, DOCX, PNG, JPEG, Markdown, HTML, and plain text. Exact per-file, per-conversation, and PDF page limits depend on the model in use. Read the limits object for that model from GET /v1/models/{version} (see Verifying limits for a configured model and Model Information). For schema details, see the Models API OpenAPI documentation.

Context Optimization Tips

Compress images and split large documents into smaller sections. Use text formats (TXT or MD) over PDF when possible for better processing. Attach only relevant files to conversations and provide document summaries for very large files.

Model-Specific Limits

Different AI models have varying capabilities and limitations that affect context processing. Each model exposes numeric caps and capability flags in the GET /v1/models/{version} response: find the object in items whose id matches the model in use, then read limits (for example context length and file size ceilings, often in bytes) and capabilities. See Verifying limits for a configured model.

Models also have response timeouts, file processing timeouts, web resource timeouts, and streaming response limits. All models include content moderation for inappropriate content, safety checks, and moderation response time limits.

Next Steps