Key findings
- Hex prices Editor seats at US$36 (Professional) or US$75 (Team) per month. It meters AI in credits: about 1 for a simple question and about 10 for a data app in Hex's own examples.
- Databricks makes Genie One and Genie Agents usage by users free through January 31, 2027, and every query runs on a pro or serverless SQL warehouse the customer pays for.
- ChatGPT Business costs US$20 per user per month billed annually and covers analysis of files and connected apps, with usage limits that vary by model and task.
- Per-seat grants in Hex reset monthly and never pool, so a few heavy users draw on add-on credits while light users leave grants unused.
- At Hex's example rates, 10 Team Explorers who each ask 5 simple questions a workday use about 1,000 credits in a 20-day month against a 100-credit grant. GILD estimates the other 900 cost about US$450 in auto top-ups.
Price AI data analysis on a month of real questions, because seat prices settle little of the bill. Agent credits per question and warehouse compute per query move the total. The data team's time curating context appears on no vendor's invoice, and it is a real cost.
Three pricing models
The three products charge for different things, so compare what a month of real questions costs. List prices are in US dollars, checked October 12, 2026.
- Hex charges US$36 on Professional and US$75 on Team per Editor per month, billed monthly. Enterprise is custom and billed annually. Monthly credit grants are 30 on Professional and 40 per Team Editor, and auto top-ups buy 50 credits for US$25.
- Databricks makes Genie One and Genie Agents usage by users free through January 31, 2027, and bills service principals. Genie Code has been pay-as-you-go since July 8, 2026. Queries run on pro or serverless SQL warehouses.
- ChatGPT Plus costs US$20 per month. Business costs US$20 per user per month billed annually, or US$25 billed monthly, and Enterprise is priced by contract. Credits extend usage on eligible plans.
Hex credits scale with each task's effort
Hex charges credits by the effort a task takes: the complexity of the request, the data and context the agent processes, and the inference it needs. Its own examples run from about 1 credit for counting customers to about 5 for a 12-month forecast and about 10 for an interactive data app. Evals, scheduled Tasks, and questions from Slack and MCP draw on the same credits. Each user can see what a task cost from the menu on its completed message.
Grants are per user and reset each cycle. A shared workspace pool funded by auto top-ups or committed credits covers anyone who runs out, and admins can cap the pool's monthly spend and each user's or group's draw.
Warehouse compute is the cost that moves with adoption
Every AI answer is a query. Hex runs queries live against the warehouse, with opt-in caching, and warns that scheduled schema refreshes query the whole schema. Genie Agents require a pro or serverless SQL warehouse and embed the creator's compute credentials so every user's questions run on it.
Opening self-serve to hundreds of people adds warehouse load that neither vendor's price sheet shows. Budget it with the warehouse owner before launch.
What users report about cost
3 of the 61 public accounts discuss what Hex costs. An operations head at a mid-market entertainment company says credit limits would stop the department's non-technical team. Some of them use about 500 credits building visuals and automations against a monthly grant of 40. A head of data wants cheaper Explorer seats, and a telecommunications analyst calls the cost fair.
Hex's price list turns that account into dollars. At US$25 per 50-credit auto top-up, the 460 credits beyond the grant that the operations head's heaviest builders need cost about US$230 per user each month, three times the US$75 Team seat. Hex's Head of Data, Katie Bauer, tells admins to read past usage in Settings and set the add-on spend limit at about 90% of expected spend. She also says to cap each user's monthly credits so one power user cannot spend the team's budget. Put those caps and an auto top-up ceiling in place before non-technical builders start. Price the first month from Hex's credit examples, then reset the limits from the usage log, which shows use by user and seat type.
Methodology
GILD read the pricing and billing pages of Hex, Databricks, and ChatGPT on October 12, 2026, and Hex's credit guidance again on October 25, 2026. User opinions come from 61 public accounts by Hex users, Jan 28, 2025 to Oct 2, 2026. Prices are list prices in US dollars. GILD's pilot estimate multiplies Hex's published per-task credit examples by a stated question volume. The per-user overage figure is GILD's arithmetic from Hex's published top-up price and one G2 reviewer's credit use. Usage figures in users' accounts are their own.
Products in this piece
Sources
- Official documentation
- 9
- Public source
- 3
- Hex pricing
- Hex docs: Credits
- Hex docs: Billing FAQ
- Hex docs: Data connections introduction
- Databricks docs: Genie
- Databricks docs: Create and manage a Genie Agent
- ChatGPT docs: Pricing
- ChatGPT use case: Analyze datasets and build reports
- G2 review from an operations head on Hex's credit limits (Jun 5, 2026)
- G2 review from a head of data on Hex's agents and Explorer seat price (Nov 24, 2025)
- G2 review from a telecommunications analyst who calls Hex's cost fair (Apr 10, 2025)
- Hex blog: Tactics for credit efficiency in Hex (Katie Bauer, May 20, 2026)