Power BI Alternatives: Power BI Competitors and Tools Like Power BI, Priced From the Source
Nine competitors with the prices their own vendors publish, plus the arithmetic Microsoft leaves out. Start with the alternative that needs no report at all: ask your database a question below.
PostgreSQL · MySQL · Snowflake · BigQuery · read-only
Ask the demo database a question:
Writing SQL… Running (read‑only)… SQL Agentsql wrote
▋
Click a question. Agentsql writes the SQL, runs it read-only, and answers.
Direct answer
The main Power BI alternatives in 2026 are Tableau ($15 per Viewer and $75 per Creator per month on Standard), Looker Studio (free, with Studio Pro at $9 per user per project), Amazon QuickSight ($3 Reader, $24 Author), Qlik Cloud ($825 a month for Standard with no per-user charge), Metabase (open source, or $90 a month billed yearly on Cloud Starter), Apache Superset (free, self-hosted), ThoughtSpot ($50 per user on Pro) and Grafana Cloud ($8.00 per active user). Almost nobody leaves Power BI to save money on a seat, because at $14.00 paid yearly it is the cheapest serious BI license on the market. People leave for three reasons Microsoft documents itself: Power BI Desktop runs only on Windows, viewers need the same paid seat as builders unless you buy an F64 capacity worth roughly 357 of them, and Copilot needs a Fabric capacity of at least F2 at $262.80 a month on top of every license you already bought.
›_ power bi competitors, priced from the source
Nine alternatives, and the shape of the bill each one sends you.
Every rate read from the vendor's own pricing page on the date in the last column. No reseller figures.
The useful comparison is not which tool draws the nicest chart. It is what each vendor decided to charge you for, because that determines whether the bill grows when your company grows. Four different billing shapes appear below: per seat (Power BI, Tableau, ThoughtSpot), split roles with an account fee (QuickSight), data volume with unlimited users (Qlik), and a platform fee plus per user (Metabase, Grafana). Pick the shape that matches how your usage will grow, then argue about features.
| Tool | Billing shape | Published entry price | The line that changes the quote | Verified |
|---|---|---|---|---|
| Tableau | Per user, per role | $15 Viewer, $42 Explorer, $75 Creator (Standard) | Billed annually on an annual contract, with at least one Creator required. Enterprise is $35, $70 and $115 for the same roles. | 10 Sep 2026 |
| Looker Studio | Free, or per user per project | $0, Studio Pro $9 per user per project | Free for creators and viewers with no seat cap. Pro multiplies by projects, so an agency with 8 clients pays 8 times. | 19 Aug 2026 |
| Amazon QuickSight | Split roles, plus an account fee | $3 Reader, $24 Author, $20 Reader Pro, $40 Author Pro | A $250 per month per-account infrastructure fee triggers the moment any Pro user or Q feature is switched on. | 2 Sep 2026 |
| Qlik Cloud | Data volume, not headcount | $300 Starter, $825 Standard, $2,750 Premium per month | Billed annually. From Standard up, Qlik's own pricing page says "No cost for additional users." | 2 Sep 2026 |
| Metabase | Platform fee plus per user | Open source $0, Cloud Starter $90/mo yearly, Pro $517.50/mo yearly | Starter includes 5 users then $65 a year each. Users of your embeds count as billable users. | 2 Sep 2026 |
| Apache Superset | Open source, self-hosted | $0 license, Apache 2.0 | The license is free. The Kubernetes cluster, the upgrade path and the engineer who owns it are not. | 2 Sep 2026 |
| ThoughtSpot | Per user, with a question cap | $25 Essentials, $50 Pro per user per month | Essentials runs from 5 to 50 users. Spotter, the AI agent, includes 25 queries per user per month on Pro. | 2 Sep 2026 |
| Grafana Cloud | Per active user, plus meters | Free 3 users, Pro from $19/mo, $8.00 per active user | Enterprise plugins, which is where Snowflake and Databricks live, reprice every active user to $55.00. | 20 Aug 2026 |
| Agentsql | Flat monthly, questions not metered | Starter $39/mo, Team $99/mo, Scale $299/mo, billed yearly | Flat pricing, questions not metered. The console on this page runs against a real database today. | 2 Sep 2026 |
The row missing from that table is Microsoft's own, and it is worth stating plainly why. Power BI Pro is $14.00 per user per month and Premium Per User is $24.00, both re-read on 2 September 2026 and both quoted only as "paid yearly"; Microsoft prints no month-to-month figure. That makes Power BI the cheapest published seat here that can both build and read: a dollar below a read-only Tableau Viewer, $61 below a Tableau Creator and $36 below ThoughtSpot Pro. If price per seat were the whole story, this page would not need to exist. The next two tables are the story. Our Power BI pricing page carries the full Fabric capacity working.
One number in that table deserves a label, and it is Tableau's. The edition cards on Tableau's Cloud and Server pricing pages show only a "Starting at" price, $15 for Standard and $35 for Enterprise, and those are Viewer seats. The role prices are published lower on the same pages, in the FAQ section: Creator $75, Explorer $42 and Viewer $15 per user per month on Standard, and $115, $70 and $35 on Enterprise, billed annually, read on 10 September 2026. So the Power BI gap depends on the seat. For a reader it is one dollar, $15 against $14.00. For a builder it is $75 against $14.00, about 5.4 times, because Power BI charges everyone the same seat and Tableau prices by role. Our Tableau pricing page walks through what that means at renewal, and Power BI vs Tableau puts the two rate cards side by side.
›_ the same team, priced fifteen ways
What 40 people, 8 of whom build, cost per year on published list prices.
Arithmetic on the rates in the table above. No negotiated discounts, no modeled assumptions.
Vendor pricing pages are built to resist exactly this calculation, which is why so few comparisons attempt it. Here is one concrete company: 40 people who need access, of whom 8 actually build things. Everything below is list price, annual, US dollars. Where a vendor charges a floor before anyone logs in, it is included. Three Microsoft configurations appear rather than one, because Microsoft's own ladder is the most interesting thing on the table.
| Tool | Per year | How that number is built |
|---|---|---|
| Looker Studio (free tier) | $0 | Free for creators and viewers, no seat cap published by Google. |
| Apache Superset | $0 | Apache 2.0. Hosting plus an engineer to run it is the real cost. |
| Metabase Cloud Starter | $3,355 | $1,080 a year covers the first 5 users, then 35 more at $65 a year each. |
| Amazon QuickSight | $3,456 | 8 Authors at $24 plus 32 Readers at $3, times 12. Add $250 a month if anyone needs Pro. |
| Agentsql Scale | $3,588 | $299 a month billed yearly, flat, questions not metered. |
| Grafana Cloud Pro | $3,780 | $19 platform fee plus 37 chargeable active users at $8.00, times 12. $24,648 with Enterprise plugins. |
| Looker Studio Pro | $4,320 | 40 users at $9 a month, for ONE project. A second project doubles it. |
| Microsoft Power BI Pro | $6,720 | All 40 seats at $14.00 paid yearly. Viewers need Pro too, so 32 of these buy read access. |
| Power BI Pro + Copilot (F2) | $9,873.60 | $6,720 of seats plus an F2 capacity at $262.80 a month, the minimum Copilot will run on. |
| Qlik Cloud Standard | $9,900 | $825 a month flat. Unlimited users, capped on data volume instead. |
| Metabase Pro | $10,110 | $6,210 a year covers the first 10 users, then 30 more at $130 a year each. |
| Power BI Premium Per User | $11,520 | 40 seats at $24.00 paid yearly. Buys bigger models and refreshes, not free viewing. |
| Tableau Standard | $12,960 | 8 Creators at $75 plus 32 Viewers at $15, times 12, on Tableau's published Standard role prices. |
| ThoughtSpot Pro | $24,000 | 40 seats at $50, each including 25 Spotter queries a month. |
| Power BI Pro (8) + Fabric F64 | $61,376 | The only Microsoft path where viewers are free: 8 Pro seats plus 64 CU at $938.00 a year each. |
Read the three Microsoft rows together and the licensing model becomes visible. Forty Pro seats is $6,720, of which $5,376 buys read access for the 32 people who will never build anything. The obvious question is whether you can buy capacity instead of seats and let those 32 read for free. Microsoft answers it in a footnote on its own capacity documentation: all SKUs other than P and F64 or larger require a Pro or Premium Per User license to consume Power BI content. So an F2, an F4, an F8, an F32, none of them free your viewers. The first one that does is F64, and at the $938.00 per capacity unit per year published in the Azure retail price API that is $60,032, before the 8 authoring seats.
Divide it out and the break-even is exact: $60,032 divided by $168 a year per Pro seat is 357.3 users. Below roughly 357 viewers, seats are cheaper. Above it, capacity is. There is nothing in between, and that discontinuity is the single most expensive thing about Power BI licensing at 40 to 300 people. QuickSight is the direct answer to it, because Amazon simply sells a $3 reader seat: 32 readers there cost $1,152 a year against $5,376 in Power BI, for the same act of looking at a chart.
The Copilot row is the other one worth arguing about. Copilot is not in the Pro seat and it is not in the Premium Per User seat; it needs a Fabric capacity of at least F2 or P1 behind it. At the $0.18 per capacity unit hour published in the Azure retail price API, an F2 is 2 x $0.18 x 730 hours, or $262.80 a month, charged whether anyone asks a question or not. That is $3,153.60 a year, and it moves Power BI from the cheapest serious option on this table to mid-pack. If plain-English questions are what you actually wanted, that capacity is the number to compare, not the seat. Our conversational analytics page normalizes what every vendor charges for the same feature.
›_ the constraint no discount fixes
Power BI Desktop is Windows only. Here is who else has that problem.
Operating system support for the authoring client, checked against each vendor's own documentation.
Every price on this page is negotiable. This is not. Microsoft's installation documentation states the requirement for Power BI Desktop as "Windows 10 or Windows Server 2016 or later", read on 2 September 2026, and there has never been a macOS build. You can view and lightly edit a published report in a browser at the Power BI service, but connecting a new data source, writing DAX measures and publishing a report all happen in Desktop, so on a Mac the authoring half of the product simply does not exist.
| Tool | Authoring client | macOS | Linux | What the documentation says |
|---|---|---|---|---|
| Microsoft Power BI | Power BI Desktop, required for authoring | No | No | Microsoft states the requirement as "Windows 10 or Windows Server 2016 or later." No macOS build exists. |
| Tableau | Tableau Desktop | Yes | No | Tableau's deployment docs document a macOS .app bundle and macOS command-line activation. The only major vendor with a native Mac authoring client. |
| Looker Studio | None, browser only | Yes | Yes | Nothing to install. Authoring and viewing both happen in the browser. |
| Amazon QuickSight | None, browser only | Yes | Yes | Authoring runs in the AWS console. There is no desktop application to install on any operating system. |
| Qlik Cloud | None, browser only | Yes | Yes | Qlik Cloud authoring is browser based. The legacy Qlik Sense Desktop client was Windows only, which is a separate product. |
| Metabase | None, browser only | Yes | Yes | Cloud is a browser app. Self-hosting ships as a JAR and a Docker image, so the server can be Linux and the analyst can be on anything. |
| Apache Superset | None, browser only | Yes | Yes | Python and Docker, usually run on Linux. Every user reaches it through a browser. |
| ThoughtSpot | None, browser only | Yes | Yes | Search and Spotter both run in the browser. Nothing to install. |
| Grafana Cloud | None, browser only | Yes | Yes | Dashboards are edited in the browser. Provisioning as code is a file in git, not a desktop app. |
| Agentsql | None, browser only | Yes | Yes | You type the question in a browser and the SQL comes back in the same window. Same on macOS, Windows, Linux, iPad. |
The two workarounds everyone reaches for are both worse than their reputation, and the evidence is on the same Microsoft page. The first is a hosted Windows desktop from a cloud provider. Microsoft's line is that "Citrix VDI and other virtual desktop environments, excluding Azure Virtual Desktop and Windows 365, aren't supported", which means the cheap third-party Windows-in-a-browser services people buy for exactly this purpose put you outside support. The second is a local virtual machine, and on Apple silicon that means Windows on ARM, for which the same page says the 2025-09 Cumulative Update (KB5065789) is required. Neither is impossible. Both mean paying for Windows licensing, a VM product and the support ambiguity, on top of the $14.00 seat.
Tableau is the honest counter-example and it deserves the credit: its own deployment documentation covers a macOS application bundle with macOS command-line activation, so Tableau Desktop is a genuine native Mac authoring client. Everything else in the table solved the problem by not having a desktop client at all. That is worth noticing as a pattern rather than a feature. The tools built after 2015 assume the browser is the application, which is also why they tend to be the ones that let a non-analyst get to a number without installing anything.
›_ choose by the problem, not the feature list
Four reasons teams replace Power BI, and what each one points to.
01
You are leaving because your team is on Macs
This is the one reason a cheaper tool does not solve, because the constraint is not money. Power BI Desktop needs Windows, and the two usual escapes are both worse than they look: Microsoft says virtual desktop environments other than Azure Virtual Desktop and Windows 365 are not supported, and Boot Camp no longer exists on Apple silicon. Tableau is the only big vendor with a native Mac authoring app. Everything else on this page runs in a browser.
02
You are leaving because viewers cost as much as builders
In a Pro workspace, reading a report needs the same $14.00 seat as building one. Microsoft's own footnote says every SKU below F64 still requires a Pro or PPU license to consume content, so a small Fabric capacity does not fix it. QuickSight is the sharpest contrast: $24 to author and $3 to read, an eight-to-one split Microsoft does not offer at any tier.
03
You are leaving because of DAX and the modeling step
Somebody has to import the data, define the relationships, write the measures and publish the report before anyone sees a number, and that somebody is usually one person the whole company queues behind. Swapping to Tableau or Qlik moves the queue, it does not remove it. The tools that remove it are the ones that take a typed question against the live schema instead of a modeled dataset.
04
You are leaving because Copilot was not what you bought
Copilot is not included in a Pro or Premium Per User seat. It requires a Fabric capacity of at least F2, which is $262.80 a month pay-as-you-go before a single question is asked. If the reason you upgraded was to ask questions in plain English, price that capacity against tools where asking a question is the product rather than an add-on.
›_ the details that do not fit in a table
What each alternative's fine print actually says.
QuickSight is the direct answer to the viewer problem
Amazon prices reading and building separately: $3 a month for a Reader and $24 for an Author, re-read on 2 September 2026. That eight-to-one split is exactly the thing Power BI does not offer at any tier, and it is why QuickSight comes out near the bottom of the 40-person table. Two triggers to model before you commit. A $250 per month per-account infrastructure fee applies once any Pro user or Q generative feature is switched on, so a 40-person deployment jumps from $288 a month to $538 the day you enable the AI. And SPICE in-memory capacity comes as 10 GB per provisioned Author, so performance is bought through the author count rather than separately. The full working is on our QuickSight pricing page.
Qlik is the only one that does not price people
Qlik Cloud lists $300 Starter, $825 Standard and $2,750 Premium per month, billed annually, re-read on 2 September 2026 and unchanged since early August. From Standard upward its own pricing page carries the phrase "No cost for additional users." That inverts the complaint that sends most people to this page. The catch sits on the other axis: Starter is capped at 10 users and a fixed 10 GB that Qlik states cannot be extended, and the higher tiers buy data capacity rather than headcount. So Qlik is expensive for a small team with a big warehouse and unusually cheap for a large team with a modest one. Worth pricing against Power BI at 100 users and again at 400. See Qlik vs Power BI for the crossover arithmetic.
Metabase counts your customers as users
Metabase's pricing page repeats one sentence under every paid tier, read again on 2 September 2026: "Both your internal team developing analytics and users of your embeds accessed through your product count as users." It matters because Metabase is the tool most often recommended as the cheap embedded analytics answer to Power BI Embedded. Cloud Starter is $100 a month, or $90 billed yearly, first 5 users included, then $6 a month or $65 a year each. Pro is $575, or $517.50 billed yearly, first 10 included, then $12 a month or $130 a year each. Put a dashboard in front of 500 customers on Starter and the arithmetic is 495 seats, not one platform fee. The open source edition stays free, which is where most teams embedding at volume end up.
Looker Studio is free until it is per project
Looker Studio is genuinely free for both creators and viewers with no published seat cap, which is the strongest free offer in this comparison and a real reason people leave Power BI for it. Studio Pro is where the pricing gets unusual: $9 per user per month PER PROJECT, and the Google SKU is literally a project subscription. Five users across eight client projects is $360 a month, not $45. Agencies get caught by that multiplier more than anyone. Note also that Looker Studio and Looker are different products with different pricing, and Google publishes no dollar price for any Looker platform edition at all. Our Looker pricing page separates the two.
Open source is free until you connect Snowflake
Apache Superset and the Metabase open source edition genuinely cost nothing to license, and for a team with platform engineers that is a real answer to per-seat Power BI. Grafana shows what the hidden edge looks like. Grafana visualization is $8.00 per active user per month, but the version with Enterprise plugins is $55.00, and the Enterprise plugin list is where Snowflake, Databricks, Oracle, MongoDB and Salesforce live. Postgres, MySQL and Prometheus are free. So the data source you connect can reprice every active user on the account, not only the person who opened the panel. The numbers are on our Grafana pricing page.
ThoughtSpot caps the thing you are buying
ThoughtSpot is the closest of the traditional vendors to a plain-English answer engine, and it is the most expensive per-seat line in the 40-person table at $24,000 a year. Read on 2 September 2026: Essentials is $25 per user per month and runs from 5 to 50 users; Pro is $50 and goes to 1,000 users. Spotter, the AI agent, is not in Essentials, and on Pro each seat includes 25 Spotter queries per month with usage past that billed on top. Twenty-five questions is roughly one per working day. That fits an executive checking a number and does not fit an analyst who is exploring, which is worth modeling before signing. Our ThoughtSpot pricing page has the credit detail.
›_ the cost nobody puts in the business case
Your .pbix files do not convert. Budget for rebuilding them.
There is no file-level converter that turns a Power BI report into a Tableau workbook or a Metabase dashboard, and the reason is structural rather than a gap somebody will close next year. A .pbix bundles exactly one semantic model plus a report definition built on it, and the logic that makes the numbers correct lives in DAX measures evaluated against that model's relationships and filter context. A Tableau workbook has no equivalent container: each worksheet binds to whichever data source it needs, and calculations evaluate in the context of a worksheet rather than a model. Nothing mechanical maps between the two directions.
Power Query is the second half of the problem and it is usually the bigger one. The M scripts in your dataflows are the actual data pipeline for most Power BI estates, and they encode business rules that exist nowhere else: which rows to exclude, how to reconcile two source systems, which date to treat as the transaction date. Moving to any other tool means rewriting that logic in SQL, in dbt, or in the target's own transformation layer, and the rewrite is where the schedule slips. Inventory the M before you inventory the reports.
The practical move is to stop treating the migration as a port. Pull the usage metrics report and find out what is actually opened. Most Power BI estates carry a long tail of reports nobody has viewed in a year, and rebuilding those is pure waste. Rebuild the ten or twenty that matter, and for the rest, ask whether the underlying need was ever a report or just a question somebody asked once. That second category is what ad hoc reporting covers, and it does not need a semantic model at all. The same logic runs in reverse for teams coming the other way, which our Tableau to Power BI migration guide goes through.
›_ a different kind of alternative
Agentsql replaces the question, not the report.
Eight of the nine tools above are the same product as Power BI at a different price. You still connect a data source, still model it, still build a view, still publish it, and still pay for the person who knows how. If the reason you are shopping is the licensing model, swapping vendors works and the tables above tell you which one to pick. If the reason is that reports get built and then nobody opens them, swapping vendors changes nothing, because the workflow is identical.
Agentsql is the other shape. It connects read-only to the PostgreSQL, MySQL, Snowflake or BigQuery database you already run, reads the live schema so it uses your real table and column names, turns a plain-English question into SQL written for that engine's own dialect, executes it and returns the number, a chart and the exact query it wrote. There is nothing to build before the first question, no DAX, no semantic model to publish, no authoring seat, and nothing to install on any operating system. It works by text to SQL, and the generated query stays visible precisely so you can check it, which is the honest answer to how accurate text to SQL really is.
Where it is the wrong tool, stated plainly. If you need governed, certified metrics that every department reports against, buy a BI platform with a semantic layer and put a person in charge of it. If your Monday morning numbers already live on a report people actually open, leave them there. If your data is scattered across a dozen SaaS applications rather than sitting in a database, the modeling problem comes first and no question-answering tool skips it. And a practical note on our own pricing: it is flat and on the page, the console at the top of this page runs against a real database right now, and there is no call to book. Our pricing page shows the intended plans in full.
›_ frequently asked
Power BI alternatives questions, answered.
What is the best alternative to Power BI?
It depends on why you are leaving. Tableau is the answer if your team is on Macs, because it is the only major vendor with a native macOS authoring app. Amazon QuickSight wins on read-heavy teams at $3 per reader. Qlik Cloud wins when headcount is the problem, because from Standard up it charges for data volume and not for people. Looker Studio wins on budget at $0.
Who is Power BI's biggest competitor?
Tableau, by a wide margin, and the price gap depends on who the seat is for. For people who only read, a Tableau Viewer is $15 per user per month on Standard and Power BI Pro is $14.00, both billed annually, so the difference is one dollar. For people who build, a Tableau Creator is $75, about 5.4 times a Power BI Pro seat. Power BI charges every user the same seat, while Tableau prices by role.
Is there a free alternative to Power BI?
Yes. Looker Studio is free for both creators and viewers with no published seat cap, which is why it appears on every list. Apache Superset and the Metabase open source edition are free to license but you host and maintain them yourself. Power BI itself has a free tier, but it cannot share content with colleagues without a Pro license somewhere in the chain.
What is the open source alternative to Power BI?
Apache Superset and Metabase are the two real ones. Superset is Apache 2.0, Python based, and built for teams who will run it on Kubernetes. Metabase has an open source edition anyone can self-host plus a paid cloud on top. Neither license costs money. Both cost an engineer, which for most US teams is more expensive than the seats they replaced.
Can I use Power BI on a Mac?
Not for building reports. Power BI Desktop is Windows only: Microsoft lists the requirement as Windows 10 or Windows Server 2016 or later, and there is no macOS build. You can view and lightly edit published reports at app.powerbi.com in Safari or Chrome. Authoring on a Mac means a Windows virtual machine, and Microsoft does not support virtual desktop environments other than Azure Virtual Desktop and Windows 365.
Why are companies moving away from Power BI?
Three reasons come up repeatedly and none of them is the seat price. Viewers pay the same $14.00 as builders unless you buy an F64 capacity. Every report needs somebody to model the data and write DAX first, so the whole company queues behind one person. And Power BI Desktop does not run on macOS, which quietly excludes design, product and executive teams that standardized on Apple hardware.
What is the cheapest alternative to Power BI?
Looker Studio at $0 if your data is in Google products or a warehouse it can reach. Among paid options for a 40-person team, Metabase Cloud Starter at about $3,355 a year and Amazon QuickSight at about $3,456 come out lowest on published list prices, roughly half the $6,720 that 40 Power BI Pro seats cost.
Is Tableau better than Power BI?
Tableau is better at exploratory visual analysis and it runs natively on macOS. Power BI is better connected to Excel, Teams and the rest of Microsoft 365, and it charges builders and viewers the same $14.00 seat, while Tableau prices by role, from $15 for a Viewer to $75 for a Creator on Standard. If your team spends its days looking for patterns visually, Tableau earns its price. If your team needs a number in a report by Friday, Power BI does that for less, especially once several people need to build.
Is Power BI Copilot free?
No, and it is not included in any seat. Copilot requires a Fabric capacity of at least F2 or P1 behind it; a Pro or Premium Per User license on its own is not sufficient. At the $0.18 per capacity unit hour published in the Azure retail price API, an F2 is $262.80 a month pay-as-you-go, billed whether anyone asks a question or not.
Does Power BI require a license for viewers?
Yes, in almost every configuration. Microsoft's capacity documentation states that all SKUs other than P and F64 or larger require a Pro or Premium Per User license to consume Power BI content. So a small Fabric capacity does not make viewing free. The first capacity that does is F64, which at the published reservation rate is roughly $60,032 a year, or about 357 Pro seats.
What is the alternative to Power Query?
For transformation inside the warehouse, dbt is the common answer, and Metabase and Superset both let you build models in SQL directly. If the goal is skipping transformation for one-off questions, a text to SQL tool queries the live tables and shows you the SQL, which removes the import-and-shape step entirely for anything you are only going to ask once.
Is there a Power BI alternative for Linux?
Every browser-based tool on this page qualifies, because there is nothing to install: Looker Studio, QuickSight, Qlik Cloud, Metabase, Superset, ThoughtSpot, Grafana and Agentsql all run in a browser on any operating system. Tableau Desktop does not ship a Linux build either, so on Linux the browser tools are the whole field.
Is Power BI being replaced by Fabric?
Power BI is not going away, but it now sits inside Microsoft Fabric and the capacity model moved with it. Practically, that means the features people most want, Copilot and free viewing, are priced through Fabric capacity units rather than through Power BI seats. Anyone budgeting Power BI in 2026 has to price the capacity, not just the licenses.
Which databases does Agentsql connect to?
Four: PostgreSQL, MySQL, Snowflake and BigQuery. Each gets SQL written for its own dialect rather than one generic string, which matters more than it sounds, because these engines disagree on date truncation, string concatenation and identifier quoting.
›_ what each one costs
›_ head to head
›_ connect your database
No DAX. No Desktop. No Windows.
Read-only on Postgres, MySQL, Snowflake or BigQuery, in any browser on any operating system. The answer comes back with the SQL beside it.