Rate Limits & Quotas
These limits protect system stability and ensure consistent performance across all users.
File Size Limits
ParseSphere enforces maximum file sizes for different operations:
Document Parsing: Files up to 200 MB
Workspace File Uploads: Files up to 200 MB
Warning
Requests exceeding these limits return a 413 status code before any
processing occurs.
{
"error_code": "file_too_large",
"message": "File too large (250.0MB). Maximum allowed: 200MB",
"details": {
"file_size_mb": 55.0,
"max_size_mb": 50
},
"request_id": "req_abc123",
"timestamp": "2025-01-03T10:30:00Z"
}Working with Large Files
For files exceeding size limits, consider compressing the document or splitting it into smaller parts before upload.
Workspace Limits
Workspaces have the following constraints:
Maximum File Size: 200 MB per file
{
"error_code": "workspace_limit_exceeded",
"message": "Workspace has reached maximum table limit (50 tables). Please delete existing tables or upgrade your plan.",
"details": {
"current_count": 50,
"maximum": 50
},
"request_id": "req_ghi789",
"timestamp": "2025-01-03T10:30:00Z"
}Chat & Query Limits
Natural language chat has the following constraints:
Maximum Iterations: 15 per message (configurable, 1-20)
Message Length: 10,000 characters maximum
Query Execution Timeout: 30 seconds per SQL query
Result Set Cap: 100 rows by default; an explicit larger LIMIT is clamped to 5,000. Scalar aggregates return 1 row.
Dataset Scoping: Up to 10 datasets per chat request
Information
SQL queries that exceed 30 seconds are terminated. The agent can make up to 15
iterations (tool calls) to answer complex questions. A query with no explicit
LIMIT returns 100 rows; a larger explicit LIMIT is clamped to 5,000.
{
"error_code": "internal_error",
"message": "Query execution timeout after 30 seconds",
"request_id": "req_jkl012",
"timestamp": "2025-01-03T10:30:00Z"
}Tip
If queries timeout, try being more specific in your question or use aggregations. The chat interface allows you to refine your question with follow-ups.
Result Caching
Parse results remain cached for efficient retrieval:
Default Cache Duration: 30 minutes
Configurable Range: 60 seconds to 86,400 seconds (24 hours)
Behavior: Subsequent requests for the same parse_id return cached results without reprocessing.
Configure Cache Duration
Override the default duration with the session_ttl parameter:
curl -X POST https://api.parsesphere.com/v1/parses \
-H "Authorization: Bearer sk_your_api_key" \
-F "file=@document.pdf" \
-F "session_ttl=300" # 5 minutesShorter TTLs reduce storage costs for high-volume integrations
Longer TTLs improve response times for frequently accessed documents
API Versioning
The current API version is v1, specified in the URL path:
# All endpoints use /v1/ in the path
https://api.parsesphere.com/v1/parses
https://api.parsesphere.com/v1/workspaces
https://api.parsesphere.com/v1/api-keysVersion Policy
Information
Version increments (v2, v3) occur only for breaking changes—removed endpoints, changed response structures, or altered parameter semantics.
Breaking changes that trigger new versions:
- Removed endpoints
- Changed response structures
- Altered parameter semantics
Non-breaking changes shipped within existing versions:
- New optional parameters
- Additional response fields
- New endpoints
Warning
Monitor the changelog for deprecation notices before version transitions. We provide advance notice for all breaking changes.
What's Next?
Learn more about API usage:
- Authentication - Manage API keys
- Error Handling - Handle limit errors
- Document Parsing - File size and processing limits
- VIVI Document Intelligence - Query execution limits