Overview
Queries
Daily chat conversations
Tokens
Monthly conversation volume
Compute
Code execution server hours
Cases
Active patient cases
Storage
Total data on disk
Users
Team members on your account
Queries
Queries are the number of chat conversations you can initiate per day. Each time you start a new conversation with Purna, it counts as one query. Continuing an existing conversation does not use additional queries. Queries reset daily at midnight UTC.Tokens
Tokens represent your total monthly chat volume. Every message you send to Purna and every response Purna sends back consumes tokens. Think of tokens as the total size of your conversations — the longer and more detailed the exchange, the more tokens it uses. A typical clinical query uses roughly 2,000 to 5,000 tokens. Complex analyses with large variant datasets or detailed reports will use more. What counts toward tokens:- Your messages and prompts
- Purna’s responses, including formatted text, tables, and citations
- Data sent to and from AI tools during a conversation (variant search results, database lookups, etc.)
Token usage resets at the start of each billing cycle. You can monitor your current usage in account settings.
Compute
Compute measures how many hours your Code Execution servers run each month. By default, every account is allocated a 4 GB, 2-core machine for custom analysis queries. When you ask Purna to write and run Python or R code, it provisions a private server attached to your chat session. Compute time is counted while that server is actively executing code. A typical analysis query takes under 4 to 5 minutes, though larger datasets or complex pipelines may take longer. How compute is metered:- Active execution time is billed per hour
- Servers automatically pause between executions and do not consume compute while paused
- Idle servers shut down after an extended period of inactivity
Cases
Cases represent the number of active patient cases you can work with in a month. An active case is one that is ready for analysis — its variants are indexed and available for querying. Lifecycle:- Cases are automatically moved to an inactive state after 90 days without activity
- You can also deactivate cases manually at any time
- Inactive cases do not count against your monthly quota but still consume storage
- Reactivating a case makes it active again and counts toward your quota
- VCF files in GRCh37 (hg19) or GRCh38 (hg38) genome builds
- All annotations are processed through our pipeline built on Ensembl VEP
If you reach your case limit, deactivate cases you are no longer actively analyzing to free up quota. Their data remains in storage and can be reactivated later.
Storage
All cases, uploaded files, and generated artifacts count toward your storage quota. Storage is measured in total disk usage across your account. Typical storage usage:| Case Type | Approximate Size |
|---|---|
| Whole Exome Sequencing (WES) | Up to 100 MB per inactive case |
| Whole Genome Sequencing (WGS) | Up to 1 GB per case, including metadata |
Inactive cases still consume storage even though they do not count against your case quota. To fully reclaim space, delete cases you no longer need.
Users
Users are the number of team members on your account. Each user gets their own login, chat history, and case access based on the permissions you configure. User management is available in Team settings. Invite new members, assign roles, and control who can access specific cases.Tools & API
All tools and API access are included in every plan, including the Free tier. This covers variant search, NCBI lookups, chart creation, code execution, gene panel search, and all other built-in capabilities. If you need a specific integration that is not currently available, contact our team at contact@purna.ai and we will work with you to set it up.Enterprise
Enterprise customers have access to additional capabilities beyond standard plans.Bring your own data sources
Bring your own data sources
Connect your own professional databases and reference data. Enterprise accounts support integration with third-party sources such as HGMD Pro or VarSome Pro using your own API keys.
FastQ file processing
FastQ file processing
Upload raw FastQ files for end-to-end processing. FastQ alignment and variant calling runs on Nvidia Parabricks GPU infrastructure for high-throughput analysis.
Custom pipeline configurations
Custom pipeline configurations
Request custom annotation pipelines, reference data sources, or processing configurations tailored to your workflow. Custom pipelines are available on request and charged separately.