Public API pricing
Get instant access to our powerful API and integrate it into your app. Check out documentation.
Small
$70/mo
- Generate SQL with natural language
- Optimize SQL Queries
- Fix SQL Queries
- Simplify SQL Queries
- Format SQL Queries
- Database Indexes
- Analyze Data
- 1,000 mo/req with RAG
- 2,000 mo/req without RAG
- 1 million AI tokens/mo included
- 50 RAG training pairs
Medium
$140/mo
- Generate SQL with natural language
- Optimize SQL Queries
- Fix SQL Queries
- Simplify SQL Queries
- Format SQL Queries
- Database Indexes
- Analyze Data
- 5,000 mo/req with RAG
- 10,000 mo/req without RAG
- 2 million AI tokens/mo included
- 100 RAG training pairs
Large
$210/mo
- Generate SQL with natural language
- Optimize SQL Queries
- Fix SQL Queries
- Simplify SQL Queries
- Format SQL Queries
- Database Indexes
- Analyze Data
- 20,000 mo/req with RAG
- 40,000 mo/req without RAG
- 5 million AI tokens/mo included
- 300 RAG training pairs
XL
$420/mo
- Generate SQL with natural language
- Optimize SQL Queries
- Fix SQL Queries
- Simplify SQL Queries
- Format SQL Queries
- Database Indexes
- Analyze Data
- 50,000 mo/req with RAG
- 100,000 mo/req without RAG
- 10 million AI tokens/mo included
- 600 RAG training pairs
FAQ
RAG stands for retrieval-augmented generation. It is used together with traditional AI prompt to teach AI how it should interpret and respond to different prompts.
RAG is used only by the text to SQL generator (textToSQL).
I recommend using the generators without RAG first and then see if adding specific RAG training instructions is needed. Usually generators perform very well without RAG training.
To use RAG you need to add a data source that you want to use it with, ie. train AI. When you have added your data source, go to the SQL generator, and generate correct prompt/code pairs and click the "Train" button to save them.
Now in the future when you generate SQL for that specific data source, we will match relevant prompt/code pair using vector search. You can use this with the API as well.
You can read more about it here.
Each time you click the "Train" button on the SQL generator and save a prompt/code training, then you create a vectorized training pair.