Understanding SmartBucket Pricing
Overview
SmartBucket operations use a token-based pricing model where costs are determined by the complexity and type of queries executed. SmartBuckets use a unified flat-rate token system i.e input token cost is equal to output token cost. Token consumption is calculated based on the underlying AI agent operations, not the final output size or format.
Core Query Types and Token Usage
Search Operations
Search operations utilize an advanced AI agent to process queries and analyze data including derived metadata. Token consumption scales with query complexity, particularly with the number of sub-questions or conditions in a query.
Base token consumption:
- Base search query: ~2,750 tokens
- Additional sub-question: ~1,100 tokens per question
Examples of token consumption:
- “Give me all my PDFs” - uses approximately 2,750 tokens
- “Give me all my PDFs without pictures of cats” - uses approximately 3,850 tokens (2,750 + 1,100)
- “Give me all my PDFs without pictures of cats, that do not contain PII” - uses approximately 4,950 tokens (2,750 + 1,100 + 1,100)
Chunk Search Operations
Chunk search operations follow a simpler token model with consistent token usage:
- Each chunk search request: ~900 tokens
- Token consumption remains constant regardless of result size
Document query
Document query operations scale with both input document size and query complexity:
- Base cost per document query: ~1000 tokens
- Document processing: ~800 tokens per page of text
- Query response generation: Varies based on requested output format and length
Token breakdown for a typical document operation:
- Base cost: 1,000 tokens
- Document processing: 4,000 tokens (5 pages × 800 tokens)
- Summary generation: ~1,000 tokens
Total: ~6,000 tokens