February 10th, 2026
PromptQL Pricing and Billing: The Prepaid Model Explained
By Drew Hahn · 17 min read
After analyzing PromptQL's pricing structure and talking to teams who use it, here’s how the prepaid credit system works and what you'll pay per program execution in 2026.
PromptQL pricing: At a glance
PromptQL pricing uses a prepaid consumption model with two main options for handling AI costs.
You can use Hasura’s managed service, where PromptQL runs on Hasura’s infrastructure and AI usage is bundled into one bill. Or, you can bring your own LLM API key, such as Anthropic, AWS Bedrock, or Google Vertex AI, and pay token costs directly to the provider.
Here's how the two options compare:
Pricing option | What you pay | What's included | Best for |
|---|---|---|---|
Hasura-managed AI service (LLM) | Per-program fee + LLM token costs (bundled into prepaid credits) | Program execution and LLM usage in one bill | Teams that want simple billing without managing multiple vendors |
Bring your own LLM key | $0.042 per program to PromptQL + token costs directly to Anthropic | Program execution only (you handle LLM costs separately) | Teams with existing Anthropic discounts or those wanting tighter control over LLM spending |
Each time PromptQL runs a program to answer your question, it costs $0.042. A program is the code PromptQL generates and executes to process your request. Simple questions might need just one program, while complex analysis can require several.
If you use Hasura’s managed LLM service, token usage is added on top of the program fee. Token pricing depends on the underlying model Hasura uses and scales with how much text the model processes in each request and response. Hasura’s managed service can run PromptQL on models from providers like Anthropic, AWS Bedrock, and Google Vertex AI.
Note: Some enterprise customers may also see DDN data transfer fees. These only apply to Private DDN Dedicated deployments with isolated cloud infrastructure, not to standard PromptQL usage. I’ll explain what these fees cover and who pays them later in this guide.
PromptQL pricing breakdown
PromptQL offers two ways to handle costs. The main difference comes down to whether you want Hasura to manage your LLM billing or handle it yourself through Anthropic. Both options charge $0.042 per program execution, but they split on who bills you for the token costs that come from processing your questions.
Here are the two options:
Option 1: Hasura-managed LLM (all-inclusive)
In this option, Hasura runs the entire infrastructure for PromptQL. This includes the LLM service that processes your natural language questions. When you choose this option, you buy prepaid credits from Hasura and those credits cover both program execution and LLM token usage.
Here’s what you need to know:
What you pay: Program fees ($0.042 per program) and LLM token costs, both bundled into your prepaid credit balance.
How it works: You purchase a set amount of credits upfront. Every time you run a query, PromptQL deducts the program cost and token usage from your balance. When your credits run low, you add more to keep using the platform.
What's included: Program execution and all LLM usage in one unified bill.
Best for: Teams that want straightforward billing without tracking costs across multiple vendors.
Pros: Single invoice, no need to manage separate Anthropic accounts, and a predictable credit-based system.
Cons: Less flexibility to negotiate LLM pricing separately, locked into Hasura's bundled rates.
Option 2: Bring your own LLM key
With option 2, you connect your own LLM API key (for Anthropic, AWS Bedrock, or Google Vertex AI) and pay PromptQL only for program execution. Token costs are billed directly by the LLM provider you choose.
Here’s what you need to know:
What you pay: $0.042 per program to PromptQL, plus the Anthropic token charges based on your contract
How it works: You still buy prepaid credits from PromptQL, but those credits only cover the $0.042 program fee. LLM token costs are billed directly by your chosen LLM provider. That means you manage two separate billing relationships.
What's included: PromptQL program execution only (you handle all LLM costs separately through Anthropic).
Best for: Teams with existing Anthropic volume discounts or those that need granular control over LLM spending.
Pros: You can leverage existing Anthropic pricing agreements, separate budget tracking for programs vs tokens, and flexibility to switch LLM providers.
Cons: Two vendors to manage, more complex accounting, and requires an existing Anthropic relationship.
What are the additional Private DDN costs for?
Private DDN costs are data transfer fees that apply when you run PromptQL on a Private DDN Dedicated deployment. These charges are infrastructure-related, not PromptQL usage fees, and only affect customers using isolated enterprise cloud setups.
DDN stands for Data Delivery Network. It’s Hasura’s managed infrastructure layer that sits between your databases and Hasura’s services, including PromptQL. When you use Private DDN Dedicated deployment, your setup runs in its own isolated space on AWS, GCP, or Azure.
Data moving between cloud regions, availability zones, or out to the internet triggers standard network egress fees from your cloud provider. Hasura passes those fees through to you without adding a markup.
Most PromptQL users won't see these costs. They only show up when all of the following are true:
You're using Private DDN Dedicated (not the standard PromptQL setup)
Your infrastructure runs in isolated cloud environments
Data moves across regions, zones, or exits to the public internet
Your cloud provider charges network egress fees for that movement
To be clear, these are standard cloud networking charges, not a PromptQL pricing tier or per-query fee. You’ll only encounter them if you’re using a Private DDN Dedicated deployment, which is reserved for customers running PromptQL in isolated enterprise cloud environments.
PromptQL pricing example
To understand what PromptQL actually costs, I looked at how pricing breaks down across different use cases. Here's an estimate of what teams can expect to pay based on query complexity:
Simple query: Single data lookup
If you ask PromptQL, "How many customers signed up last month?" here’s what you can expect:
Programs executed: 1
PromptQL fee: $0.042
LLM token usage: ~2,000 tokens (basic question, short response)
LLM cost: ~$0.006 (at Anthropic's Sonnet pricing)
Total cost: ~$0.05 per query
Complex analysis: Multi-step workflow
Let’s say, you ask PromptQL: "Show me revenue trends by product category over the last 6 months, then identify which categories are declining and explain why." You can expect:
Programs executed: 3-5 (retrieve data, analyze trends, generate explanations)
PromptQL fee: $0.126 - $0.21 (3-5 programs × $0.042)
LLM token usage: ~8,000-10,000 tokens (complex prompt, detailed analysis, longer output)
LLM cost: ~$0.024 - $0.030
Total cost: ~$0.15 - $0.24 per query
High-volume monthly usage
A team running 500 mixed queries per month uses a mix of simple and complex queries. Here's what that costs:
Average cost per query: ~$0.08
Monthly total: ~$40
If you scale that up to 2,000 mixed queries per month, it would cost:
Average cost per query: ~$0.08
Monthly total: ~$160
Is PromptQL worth the cost?
PromptQL lets you ask questions about your data in plain English instead of writing SQL. The pricing works well for teams that can estimate their monthly queries. It’s also suitable for those who don’t mind buying credits ahead of time.
Here's how to decide:
It's worth it if: You want quick answers from your data without learning SQL and you run a steady number of queries each month. Once you know your usage, costs stay predictable.
It's best for: Teams that pull data from multiple databases who want one tool to handle connections, query building, and AI processing without building it themselves.
Skip it if: You need to see pricing before calling sales, you run thousands of queries per month where per-query costs add up fast, or your team already knows SQL and prefers working directly with databases.
Overall, I see PromptQL as a good fit for business teams that need quick answers from their data to support revenue and performance decisions. The main thing to watch is usage volume. Costs rise with every query, and at higher volumes, per-query pricing can outpace flat-rate tools.
PromptQL alternatives and pricing comparison
Tool | Starting price (billed annually) | Best for | Key advantage |
|---|---|---|---|
Business users who want quick insights without SQL | Natural-language analysis with charts, insights, and repeatable notebooks | ||
$36/editor/month, billed monthly | Technical teams that want collaborative notebooks | SQL, Python, and AI combined in one workspace with version control | |
Data teams that need real-time collaboration | Notebook environment with live editing and built-in integrations |
Julius: Best for business users who want quick insights without SQL
We designed Julius for business users who need quick data insights without learning SQL or Python. You upload your data as CSV or Excel files, or connect databases like Postgres, BigQuery, and Snowflake. Then, you can ask questions in plain English to get analysis, charts, and reports.
The platform combines chat-based exploration with structured Notebooks for repeatable analysis, so you can save important questions and return to them as your data updates. For example, a saved revenue or performance check can refresh automatically when you rerun or schedule the notebook.
Julius learns your database structure over time by understanding how your tables connect and what your columns mean. This helps Julius return more consistent answers by reusing those learned relationships instead of reinterpreting your data from scratch each time.
Julius starts at $37per month.Hex: Best for technical teams that want collaborative notebooks
Hex is a collaborative data workspace that combines SQL, Python, and AI in one platform. I tested how its interface handles complex queries. I found the platform is built for data analysts and engineers who want version control, parameterized reports, and team collaboration on analysis workflows.
Hex pricing starts at $36 per editor per month, billed monthly. It fits teams that have SQL and Python skills and need a shared environment for building and sharing data work.Deepnote: Best for data teams that need real-time collaboration
Deepnote is a collaborative notebook platform where multiple team members can work on the same analysis at once. I tested how its live editing works and found that the system is designed for data teams that want Google Docs-style collaboration on queries, visualizations, and documentation.
Deepnote pricing starts at $39 per editor per month. It's a good fit for teams that want real-time editing and built-in connections to common data sources without managing separate infrastructure.Julius vs PromptQL: Which should you choose?
Julius covers the full data analysis experience with file uploads, database connections, visualizations, notebooks, and scheduled reports in one system. PromptQL focuses on backend AI programs that query data, while analysis, visualization, and reporting happen elsewhere.
Use this guide to figure out which one fits your work:
Julius is better for: Business users who want straightforward data analysis without SQL skills. It's a strong fit if you need charts, repeatable Notebooks, and scheduled reports in one place without managing infrastructure or prepaid credits.
PromptQL is better for: Teams already using Hasura's data infrastructure who want to add natural language queries on top. PromptQL pricing works well if you can estimate query volume upfront and prefer Hasura to manage database connections and LLM services together.
Use both if: You want Julius for business user analysis and reporting, while using PromptQL for backend data access through Hasura's infrastructure.
Want to see how Julius can simplify data analysis and reporting? Try Julius for free today.
My bottom line on PromptQL pricing
I think PromptQL makes the most sense for teams already using Hasura who want to add natural language queries on top of their existing data stack. Once usage patterns are clear, the per-program pricing becomes predictable, and the prepaid credit model works well for steady query volumes.
Where I see the tradeoff is in upfront cost modeling. Because pricing depends on how many programs each question triggers, I found it harder to estimate spend early on, especially during evaluation or periods of spiky usage. At higher volumes, those per-program fees can also grow faster than flat monthly pricing.
If pricing transparency and easier forecasting matter more to you, I’d recommend trying Julius instead. Julius offers clear monthly plans without per-query charges, which makes it easier to budget as usage grows.
Frequently asked questions
Does PromptQL offer a free trial?
PromptQL does not offer a fully self-serve free trial with instant sign-up. To access the product, pricing details, or a live demo, you need to engage with the PromptQL sales team. While guided demos, pilots, or limited playground access may be available, there’s no open trial where you can independently test real workloads before committing to prepaid credits.
What's included in the per-program fee?
The per-program fee covers the code generation and execution that PromptQL runs to answer your question. Each time you ask something, PromptQL writes and runs a program to pull data from your sources and return results. Simple queries need one program, while complex analysis can trigger multiple programs in a single request.
How do prepaid credits work?
Prepaid credits work like a balance you load upfront that gets deducted each time you run a query. You buy a set amount of credits from PromptQL. Then, each program execution costs $0.042 plus any LLM token fees if you use Hasura's managed service. When your balance runs low, you add more credits to keep using the platform.
