Cloud Limits

Default Quotas, Rate Limits, & File Size Constraints

SciPhi Cloud imposes limits to govern the maximum number of documents, chunks, and collections per user, as well as rate limits for requests and file-upload sizes. The same defaults apply when you deploy R2R locally (though you can override them in your local r2r.toml).


Overview of Limits

Default Limits

Below are the default per-user limits for all R2R Cloud deployments:

Resource / ActionDefault LimitNotes
Documents100Maximum number of documents you can upload
Chunks10,000Maximum total chunks derived from uploaded files
Collections5Maximum number of collections per user
File Upload (global)2 MB (2,000,000 bytes)Overall max file upload size if no extension-specific limit
File Upload (by extension)See table belowCertain file types can have their own max size
Rate Limit (global)60 requests/min per user-

Note: Additionally, our Nginx ingress layer applies a 60 requests/minute per IP limit.
Thus, a single IP cannot exceed 60 requests/minute, regardless of user account.


File Upload Size by Extension

R2R Cloud enforces specific maximum file sizes for different file types. If no extension-specific limit is found, we fall back to the global 2 MB limit.

ExtensionMax Size (bytes)Approx. MBNotes
txt2,000,000~2.0Plain text files
md2,000,000~2.0Markdown
csv5,000,000~5.0CSV spreadsheets
xml2,000,000~2.0XML markup
doc10,000,000~10.0MS Word (legacy)
docx10,000,000~10.0MS Word (modern)
ppt20,000,000~20.0MS PowerPoint (legacy)
pptx20,000,000~20.0MS PowerPoint (modern)
xls10,000,000~10.0MS Excel (legacy)
xlsx10,000,000~10.0MS Excel (modern)
pdf30,000,000~30.0PDF can expand significantly in text
epub10,000,000~10.0E-book format
jpeg / jpg / png5,000,000~5.0Images
…morevaries-Configurable in r2r.toml

You can customize or override these limits in your own r2r.toml if deploying locally.


Route-Specific Rate Limits

In the cloud.toml example, certain endpoints have additional per-route limits. For instance:

RouteRequests/MinMonthly LimitNotes
/v3/retrieval/search103,000Searching your knowledge base
/v3/retrieval/rag5200Retrieval-Augmented Generation
/v3/documents/create10200Document ingestion
/v3/retrieval/agent5100Interactive agent calls
/v3/retrieval/completions00Disabled in this sample
/v3/retrieval/embeddings00Disabled in this sample

These per-route limits are enforced in addition to the global 60 req/min user limit.
If a per-route limit is lower than the global limit, that route’s usage will be throttled sooner.


Starter Tier Overrides

When upgrading to the Starter Tier, these default limits increase:

Resource / ActionNew Limit
Documents1,000
Chunks100,000
Collections50

All other limits (like file size and request rates) remain the same unless otherwise specified by your plan.


Local Deployment

If you deploy R2R on your own infrastructure, the same default limits apply out of the box. However, you can easily override them in your local r2r.toml or equivalent configuration. For example:

1[app]
2default_max_documents_per_user = 200
3default_max_chunks_per_user = 50_000
4default_max_collections_per_user = 20
5default_max_upload_size = 10_000_000 # 10 MB

And for route-based limits:

1[database.route_limits]
2"/v3/retrieval/search" = { route_per_min = 50, monthly_limit = 10_000 }

Additional Notes

  • User-Level Overrides: Admins can grant custom overrides to specific users. For example, a single user’s “max_documents” might be raised to 5,000.
  • Monthly Limits: Resets on the 1st of each month.
  • Request Logging: R2R logs usage for each route to track compliance with monthly or per-minute limits.
  • Nginx Rate Limit: Cloud deployments also apply a 60 requests/minute limit per IP at the ingress layer.

Conclusion

R2R Cloud’s default usage limits keep the platform performant and equitable. When you need more capacity—such as uploading larger documents or making more requests—upgrading your tier or adjusting local r2r.toml (for self-hosted deployments) is straightforward.

By keeping usage within your plan’s limits and splitting large files where necessary, you’ll ensure a smooth experience with R2R’s ingestion, retrieval, and advanced AI features.

Was this page helpful?
Built with