Best BI tools for dbt in 2026: 12 options compared on dbt Core support, the Semantic Layer, and metric duplication

The best BI tools for dbt in 2026 are Lightdash and Omni if you want the BI layer to read your dbt project directly (both work with dbt Core), and Steep, Hex, or Tableau if you are on the dbt platform and want every metric defined once in the dbt Semantic Layer. Sigma and ThoughtSpot sync dbt descriptions and models into their own layers, while Metabase and Basedash query the tables dbt builds. The deciding questions are whether you run dbt Core or the dbt platform, where your metrics live today, and whether people who don’t write SQL need to explore the result. Every price and plan below was checked on the vendor’s official page on October 19, 2026.

This guide is for analytics engineers and data teams who run dbt and are choosing what sits on top of it. It covers the four ways BI tools connect to dbt, what changed in the dbt ecosystem this year, a comparison table, a fact card for each tool, and which tool fits which stack.

What does “dbt-native” mean for a BI tool?

Vendors use “dbt integration” for very different things, and the difference decides whether you maintain one metric definition or two. When you evaluate a tool, find out which of these four patterns it uses.

1. Read the dbt project directly. The BI tool connects to your dbt repo, compiles or reads manifest.json, and turns models into explorable tables. Metrics and dimensions are defined in dbt YAML. Lightdash and Omni work this way, and both work with dbt Core. The cost is that the BI experience depends on a maintained dbt project.

2. Query the dbt Semantic Layer over its API. The BI tool sends a request such as “revenue by month for the last year” to the dbt Semantic Layer through its JDBC, ADBC, or GraphQL API, and MetricFlow generates the SQL. The metric logic never leaves dbt. Steep, Hex, Tableau, Power BI, Mode, and Sigma (in preview) are among the integrations dbt lists. The Semantic Layer APIs require a paid dbt platform plan.

3. Sync dbt metadata, then define metrics in the BI tool. The tool imports descriptions, tests, keys, and sometimes models from dbt artifacts, but metrics are built in its own layer. Sigma, ThoughtSpot, and Metabase (through the community dbt-metabase tool) fall here. Documentation stays consistent, but metric logic can drift because it lives in two places.

4. Query the tables dbt built. The tool connects to the warehouse and reads the models dbt materialized. Looker and Power BI add their own modeling language (LookML, DAX) on top; Basedash reads warehouse descriptions as AI context and adds its own models. If your canonical metrics are already pre-computed in wide mart tables, this is not duplication: the metric still lives in dbt and the BI tool reads a column.

What changed in the dbt ecosystem in 2026?

Four changes affect BI tool choice this year:

• Fivetran and dbt Labs merged. The all-stock merger announced in October 2025 closed on June 1, 2026. dbt Cloud is now called the dbt platform in dbt’s docs.
• MetricFlow is open source. dbt Labs relicensed MetricFlow under Apache 2.0 in October 2025, from version 0.209.0. You can define and query metrics locally without a dbt platform plan, but the hosted APIs and BI integrations still need one.
• The Fusion engine became dbt Core v2. The Rust-based engine is now the default “v2” of dbt, and its runtime was open-sourced as dbt Core v2.0 (alpha). Lightdash documents Fusion support; ThoughtSpot’s dbt sync supports dbt 1.7 and earlier.
• The Open Semantic Interchange became Apache Ossie. The vendor-neutral semantic model spec entered the Apache Incubator as Apache Ossie in July 2026. dbt says production readiness of its Power BI connector depends on Microsoft’s participation in Ossie, and Lightdash lists an Ossie integration as under development.

How we evaluated BI tools for dbt

We checked each tool against the questions dbt-first teams ask when they pick a BI layer:

1. dbt integration pattern. Does it read the project, query the Semantic Layer, sync metadata, or only read tables? Can it write changes back to dbt as a pull request?
2. dbt Core or dbt platform. Does the integration work with dbt Core, or does it need a paid dbt platform plan?
3. Metric duplication. After setup, where does each metric definition live: in dbt, in the BI tool, or both?
4. Non-technical access. Can operations, finance, or sales teammates explore dbt models without SQL, through a point-and-click builder or natural language?
5. Governance. Row-level security, SAML SSO, and SCIM, and on which plan.
6. Price and deployment. Pricing model and starting price, free tier or trial, and cloud or self-hosted.

BI tools for dbt comparison table

“Works with dbt Core” means the dbt integration itself runs without a paid dbt platform plan. Any of these tools can query the tables a dbt Core project builds.

Lightdash

Best for: dbt-first teams that want BI defined as code in the same repo as their models, including teams on dbt Core.

Lightdash is the most directly dbt-native option. Its dbt integration reads a connected dbt project and builds the semantic layer from it: columns become dimensions, descriptions show up as hover text, and metrics live in YAML next to the models. The CLI compiles the project locally and pushes manifest.json. A write-back feature lets people develop models and metrics in Lightdash and open a pull request against the dbt project on GitHub, GitLab, or Bitbucket Cloud.

Fact card: Lightdash

• Pricing (as of October 2026): Open Source is free to self-host. Cloud Pro is $3,000 per month with unlimited users and no per-seat pricing. Enterprise is custom (pricing).
• Free tier or trial: free open source edition; 21-day Cloud trial that starts when you connect a dbt project, no payment details.
• Deployment: Lightdash Cloud, or self-hosted with Docker Compose or Kubernetes.
• dbt support: dbt Core, the dbt platform, and dbt 1.4 and later, including Fusion. MetricFlow support (beta) reads semantic models from the compiled manifest with no dbt Semantic Layer API needed, and translates simple metrics only.
• Warehouses: BigQuery, Snowflake, Redshift, Databricks, Postgres, Trino, ClickHouse, Athena, and DuckDB.
• Query model: live queries against the warehouse; cached results and pre-aggregates on Cloud Pro and above.
• Row-level security: user attributes in sql_filter and related model properties.
• SSO and SCIM: SAML SSO and SCIM 2.0 on Enterprise; Cloud Pro includes Google SSO.
• AI features: AI agents answer natural language questions from the semantic layer, and an MCP server is available to all Cloud users (self-hosted needs Enterprise).
• Embedding: JWT-scoped iframes or a React SDK as an add-on: pay as you go (first 1,000 loads free, then $0.05 per load) or $790 per month for 100,000 loads.
• Not ideal for: teams without a well-maintained dbt project, or small teams that won’t self-host and find $3,000 per month steep.

For a closer look, see Basedash vs Lightdash and Lightdash alternatives.

Omni

Best for: teams that want a governed BI model synced both ways with dbt, plus a spreadsheet-style and point-and-click experience for business users.

Omni connects to your dbt Git repository, compiles the project into a manifest, and reads it. It supports dbt Core and the dbt platform. Omni is the strongest at writing back: analysts can promote an Omni query into a dbt model, push field descriptions into schema.yml, and push dashboards to dbt as exposures, all through a pull request. Its dbt Semantic Layer integration is a one-way import that maps MetricFlow semantic models and metrics into the Omni model; cumulative and conversion metrics are not supported.

Fact card: Omni

• Pricing (as of October 2026): not published; quote only.
• Free tier or trial: trial through a scheduled session with Omni’s team.
• Deployment: cloud, hosted on AWS and Azure. Self-hosting is not documented.
• dbt support: dbt Core and the dbt platform, including dbt 2.0 version pins. Write-back requires a direct Git connection.
• Warehouses: Snowflake, BigQuery, Databricks, Redshift, Postgres, and others.
• Query model: live queries with results cached for 6 hours by default and aggregate awareness.
• Row-level security: access_filters based on user attributes.
• SSO and SCIM: SAML and SCIM for Microsoft Entra, Okta, and Rippling. Plan tiers are not documented.
• AI features: Omni Agent answers questions in plain language; an MCP server connects external AI tools to the Omni model. AI usage is metered in credits with overage billing.
• Embedding: signed iframe URLs and an Embed SDK.
• Not ideal for: teams that need published pricing or self-hosting, or teams that want MetricFlow to stay the only metric engine at query time.

See Lightdash vs Omni for the two project-reading tools side by side.

Steep

Best for: teams on the dbt platform who have adopted MetricFlow and want a metrics-first front end where dbt stays the only metric engine.

Steep is built around metrics rather than tables. With its dbt integration, Steep does not use its own semantic layer; it uses the dbt platform for definitions, query generation, and execution, and reads config.meta properties for categories, owners, and visibility. That makes it one of the few tools where defining a metric once in dbt is the only option, not just a recommended setup.

Fact card: Steep

• Pricing (as of October 2026): Pro $15 per seat per month, free up to 3 seats; Business $25 per seat per month; Enterprise custom (pricing).
• Free tier or trial: free for up to 3 seats on Pro; Business trial.
• Deployment: cloud, with web, Mac, and mobile apps.
• dbt support: the dbt Semantic Layer API, which requires a paid dbt platform plan. Steep can also sync Cube or its own metrics-as-code. dbt Core is not documented.
• Warehouses: Athena, BigQuery, Databricks, MySQL, PostgreSQL, Redshift, Snowflake, SQL Server, and Synapse.
• Query model: live queries in the warehouse with a one-hour default cache.
• Row-level security: access policies on modules, on Enterprise.
• SSO and SCIM: SAML SSO on Business and above; SCIM on Enterprise.
• AI features: Steep Agent finds relevant metrics and summarizes results; a hosted MCP server.
• Embedding: REST API for querying metrics; iframe or SDK embedding is not documented.
• Not ideal for: dbt Core teams, or teams whose analysis needs go beyond predefined metrics into ad hoc SQL.

Hex

Best for: analysts and data scientists who work in SQL and Python notebooks and want dbt metadata and metrics inside them.

Hex has two dbt integrations: a metadata integration that brings dbt metadata into Hex through the dbt Discovery API, and a Semantic Layer integration that Hex now labels legacy. Hex says neither works with dbt Core, because both rely on dbt platform features. The newer path imports dbt MetricFlow files into Hex semantic models, which Threads uses to answer natural language questions.

Fact card: Hex

• Pricing (as of October 2026): Community free; Professional $36 per editor per month; Team $75 per editor per month; Enterprise custom (pricing).
• Free tier or trial: free Community plan; 14-day Team trial with no card.
• Deployment: cloud (multi-tenant, EU, or single-tenant add-on on Enterprise).
• dbt support: the dbt platform for metadata and the legacy Semantic Layer integration; MetricFlow file import into Hex semantic models.
• Query model: live warehouse queries with a 60-minute default cache.
• Row-level security: user attributes in signed embeds; OAuth database connections on Enterprise.
• SSO and SCIM: SSO through OIDC only, SAML is not supported, on Enterprise; directory sync on Enterprise.
• AI features: Threads for natural language questions on every plan, the Hex agent, and an MCP server on Team and Enterprise.
• Embedding: signed iframe URLs with the Embedded Analytics add-on on Enterprise.
• Not ideal for: teams that require SAML, dbt Core users who want the metadata integration, or business users who expect a dashboard-first tool.

Sigma

Best for: Snowflake or Databricks teams whose business users think in spreadsheets and should see dbt documentation and test results where they work.

Sigma’s dbt integration shows dbt table and column descriptions, dbt test results, and last run times inside Sigma. It works with the dbt platform, or with dbt Core (beta) when you upload manifest.json and run_results.json from your runs. Separately, Sigma can query the dbt Semantic Layer through its JDBC API, which requires the dbt platform and is listed as a preview by dbt.

Fact card: Sigma

• Pricing (as of October 2026): quote only.
• Free tier or trial: 7-day free trial.
• Deployment: cloud, hosted on AWS, Azure, or Google Cloud.
• dbt support: metadata and tests from dbt Core (beta) or the dbt platform; Semantic Layer queries on the dbt platform only.
• Query model: live SQL against the warehouse, with browser and warehouse caching and optional materialization; Sigma does not store your data.
• Row-level security: rules based on user attributes, team membership, or email; warehouse RLS is respected on OAuth connections.
• SSO and SCIM: SAML and SCIM (SCIM requires SAML). Plan tiers are not documented.
• AI features: Sigma Assistant for natural language queries, agents in workbooks, and an MCP server.
• Embedding: JWT-secured iframes and a React Embed SDK.
• Not ideal for: teams that need self-hosting or published pricing, or teams that want metrics defined only in dbt (Sigma metrics are usually built in Sigma).

Tableau

Best for: existing Tableau shops on the dbt platform that want governed dbt metrics in Tableau without rebuilding them as calculated fields.

The dbt Semantic Layer connector for Tableau works in Tableau Desktop 2021.1 and later, Tableau Server, and Tableau Cloud. It requires live connections; extracts, custom SQL, and cross-database joins are not supported through the connector. Outside the connector, Tableau reads dbt-built tables like any other warehouse source.

Fact card: Tableau

• Pricing (as of October 2026): Standard edition Creator $75, Explorer $42, and Viewer $15 per user per month, billed annually; Enterprise edition and Tableau+ cost more (pricing). All plans need an annual contract.
• Free tier or trial: free Tableau Desktop Free Edition (no publishing); Tableau Cloud trial.
• Deployment: Tableau Cloud or self-hosted Tableau Server.
• dbt support: the Semantic Layer connector, which needs a paid dbt platform plan.
• Query model: live connections or extracts for normal sources; live only through the dbt connector.
• Row-level security: user filters, data source filters, or virtual connection data policies (Data Management add-on).
• SSO and SCIM: SAML 2.0; SCIM on Tableau Cloud.
• AI features: Tableau Pulse and Tableau Agent, which needs Tableau+ or Tableau Cloud+.
• Embedding: Embedding API v3 with Connected Apps (JWT).
• Not ideal for: dbt Core teams that want governed metrics, or small teams that don’t want annual per-user contracts.

Power BI

Best for: Microsoft-centric companies that already license Power BI and want to try dbt metrics in reports.

The dbt Semantic Layer connector for Power BI is in preview as of October 2026. It runs in Power BI Desktop and reaches the Power BI service only through an on-premises data gateway, because the service doesn’t natively support custom connectors. It is DirectQuery only, with no Import mode, custom DAX, or joins. dbt says production readiness depends on Microsoft’s participation in Apache Ossie. Most dbt teams on Power BI today point it at dbt-built tables and define measures in DAX.

Fact card: Power BI

• Pricing (as of October 2026): Pro $14 and Premium Per User $24 per user per month, paid yearly (pricing).
• Free tier or trial: free Power BI Desktop and free account.
• Deployment: Power BI service (cloud).
• dbt support: Semantic Layer connector in preview (dbt platform required); otherwise DAX on dbt tables.
• Query model: Import or DirectQuery for normal sources; DirectQuery only through the dbt connector.
• Row-level security: DAX roles, which only restrict users with Viewer permissions.
• SSO and SCIM: Microsoft Entra ID; users are provisioned through Entra.
• AI features: Copilot on F2 or larger Fabric capacity or P1 Premium.
• Embedding: embed for your organization, or for customers on paid capacity.
• Not ideal for: dbt-first teams that want one metric definition in production today, or teams outside the Microsoft stack.

ThoughtSpot

Best for: enterprises that want search and AI agents over dbt models, with ThoughtSpot models generated from dbt instead of built by hand.

ThoughtSpot’s dbt sync reads manifest.json and catalog.json and creates ThoughtSpot Models (formerly Worksheets) from your dbt models. It connects to the dbt platform with an API key, or to dbt Core through a zip upload of the artifacts. For Snowflake connections it also imports semantic definitions such as formulas, measures, dimensions, and metrics. The sync supports dbt 1.7 and earlier.

Fact card: ThoughtSpot

• Pricing (as of October 2026): Essentials from $25 per user per month (annual, 5 to 50 users); Pro $50 per user per month (annual) or usage-based from $0.10 per credit; Enterprise custom (pricing).
• Free tier or trial: free trial.
• Deployment: ThoughtSpot Cloud, or self-managed ThoughtSpot Software.
• dbt support: dbt Core (zip upload) and the dbt platform, dbt 1.7 and earlier.
• Query model: live queries; optional in-memory cache add-on.
• Row-level security: group-based rules on all Analytics plans.
• SSO and SCIM: SAML on Pro and Enterprise; SCIM supported (plan not documented).
• AI features: Spotter AI agents (25 queries per user per month on per-user Pro and Enterprise); a Spotter MCP server add-on on Enterprise and Embedded.
• Embedding: Visual Embed SDK, React components, and REST API.
• Not ideal for: teams on dbt 1.8 or later (including Fusion) who want the dbt sync, or small teams that want a short buying process.

Looker

Best for: teams committed to LookML as their semantic layer, using dbt for transformation only.

Looker is not on dbt’s Semantic Layer integration list, and Looker’s docs don’t describe consuming dbt metrics. Metrics live in LookML on top of dbt-built tables, which is a deliberate two-layer setup that many teams run. The one documented dbt link is a continuous integration trigger (August 2026) that runs Looker CI suites when a dbt platform CI job finishes, so model changes that would break Explores surface before merge.

Fact card: Looker

• Pricing (as of October 2026): quote only for Standard, Enterprise, and Embed editions (pricing).
• Free tier or trial: trial instances available.
• Deployment: Google Cloud.
• dbt support: queries dbt tables; CI trigger from dbt platform jobs.
• Query model: live SQL against the warehouse.
• Row-level security: access_filter with user attributes.
• SSO and SCIM: SAML 2.0; SCIM in preview through Workforce Identity Federation.
• AI features: Gemini in Looker, including Conversational Analytics, data agents, and LookML generation (preview).
• Embedding: private and signed embedding, including Conversational Analytics.
• Not ideal for: dbt-first teams that want one definition per metric, since LookML becomes a second metrics layer.

Metabase

Best for: teams that want low-cost or open source self-serve dashboards where dbt provides documentation, not the metric source.

Metabase has no native dbt integration. The community dbt-metabase tool reads a compiled manifest.json, so it works with dbt Core, and propagates model and column descriptions, table relationships, and semantic types into Metabase. It can also extract Metabase questions and dashboards back into dbt as exposures. Metrics themselves are usually rebuilt in Metabase.

Fact card: Metabase

• Pricing (as of October 2026): Open Source free; Starter $100 per month for 5 users plus $6 per extra user; Pro $575 per month for 10 users plus $12 per extra user; Enterprise from $20,000 per year (pricing).
• Free tier or trial: free open source edition; 14-day trial on Cloud Starter and Pro.
• Deployment: Metabase Cloud or self-hosted (Starter is cloud only).
• dbt support: community dbt-metabase sync; no Semantic Layer support.
• Query model: live queries with caching (granular and preemptive caching on Pro and Enterprise).
• Row-level security: row and column security on Pro and Enterprise.
• SSO and SCIM: SAML and SCIM on Pro and Enterprise.
• AI features: Metabot for natural language charts and SQL, using your own API key or Metabase’s AI service at $3.75 per million tokens; an MCP server.
• Embedding: guest embeds on all plans; modular embedding SDK for React on Pro and Enterprise.
• Not ideal for: teams that want metrics defined only in dbt, or that don’t want to maintain a community sync tool.

Cube

Best for: teams that want one headless semantic layer serving several BI tools, notebooks, and AI agents at once.

Cube is a semantic layer first and a BI tool second. Its dbt integration clones and parses your dbt repo or accepts the manifest.json from your CI job, and turns dbt models into cubes. Metrics then live in Cube, which serves them through its APIs to downstream BI tools, notebooks, and agents. For dbt teams, Cube is an alternative to the dbt Semantic Layer rather than a consumer of it.

Fact card: Cube

• Pricing (as of October 2026): Free; Starter $40 per developer per month; Premium $80 per developer per month with Explorer seats at $40 and Viewer seats at $20; Enterprise custom (pricing). The dbt integration needs Premium or above.
• Free tier or trial: free plan with 1,000 requests a day.
• Deployment: Cube Cloud, bring-your-own-cloud on Enterprise, or open source Cube Core.
• dbt support: dbt Core or the dbt platform through the manifest.
• Query model: live queries with pre-aggregations.
• Row-level security: access policies for row-level and member-level security and masking.
• SSO and SCIM: Enterprise.
• AI features: analytics chat, a semantic model agent, and an MCP server for AI agents.
• Embedding: embedded dashboards and analytics chat on Premium and above.
• Not ideal for: teams that already adopted MetricFlow, or small teams that want one tool rather than a semantic layer plus a front end.

Basedash

Best for: dbt teams whose canonical metrics are materialized as mart tables, and who want people who don’t write SQL to ask questions in plain English.

Basedash connects to the warehouse where dbt builds your tables and queries those tables and views in place. Table and column descriptions in the warehouse, including the ones dbt writes there with persist_docs, are read automatically as AI context. Basedash does not parse dbt project files or query the dbt Semantic Layer API. Teams that need measures and segments on top of their marts define them in Basedash Models: SQL datasets with measures, segments, relationships, a default time column, version history, and admin verification, which every Basedash AI and the MCP server reuse. Models can be managed from CI through the REST API.

Fact card: Basedash

• Pricing (as of October 2026): Startup $1,000 per month plus AI usage, with up to 25 users and $1,000 of monthly AI credits; Enterprise custom (pricing).
• Free tier or trial: 14-day trial with no credit card.
• Deployment: cloud; self-hosting on Enterprise.
• dbt support: queries dbt-built tables and reads persisted descriptions; no dbt project parsing or Semantic Layer API.
• Warehouses: Snowflake, BigQuery, Databricks, Redshift, ClickHouse, and PostgreSQL, among others.
• Query model: live SQL against the connected warehouse.
• Row-level security: Postgres only, through policies that read the basedash.groups session variable; other sources use group-based data source access.
• SSO and SCIM: SAML SSO and SCIM on Enterprise.
• AI features: an AI data analyst that writes SQL and builds charts and dashboards from questions, Insights, automations, a Slack app, and an MCP server on every plan.
• Embedding: iframe embeds and a React SDK on Enterprise.
• Not ideal for: teams that want metrics defined only in dbt YAML or MetricFlow and queried live through the Semantic Layer, or Snowflake and BigQuery teams that need row-level security enforced inside the BI tool.

dbt Semantic Layer vs a BI tool’s built-in semantic layer

A semantic layer is where metric definitions, dimensions, and relationships are centralized so every dashboard computes the same number the same way. The dbt Semantic Layer puts it in dbt. A BI tool’s built-in layer (LookML, an Omni model, Basedash Models) puts it in the BI tool.

If you want metrics defined once and used from a notebook, a spreadsheet, and a dashboard, the dbt Semantic Layer wins because the definition isn’t tied to one front end. If you use one BI tool and your analysts need to create metrics quickly without pull requests, a built-in layer is faster day to day, at the cost of a second place where logic lives. The trap is the accidental middle: defining metrics in dbt and then redefining them in the BI tool because the integration only synced descriptions. That is the metric drift problem behind most “why do these two dashboards disagree” tickets. Our semantic layer tools comparison covers standalone layers in more depth.

Which BI tool should you choose for dbt?

If you run dbt Core and want the BI tool to read your models directly: Lightdash and Omni both parse dbt Core projects, and both open pull requests back to dbt. ThoughtSpot and Sigma (beta) accept uploaded dbt Core artifacts. The Semantic Layer tools (Steep, Hex’s metadata integration, Tableau, Power BI) need a paid dbt platform plan.

If you are on the dbt platform with BigQuery or Snowflake and have adopted MetricFlow: Steep, Tableau, and Hex query or import MetricFlow definitions so metrics stay in dbt. Omni imports them into its own model. Check that the metric types you use are supported; Omni skips cumulative and conversion metrics, and Lightdash translates simple metrics only.

If an operations or finance team that doesn’t write SQL needs to explore dbt models: Omni and Sigma offer spreadsheet-style and point-and-click exploration on dbt models. Basedash, ThoughtSpot Spotter, and Metabase’s Metabot answer plain-English questions. Lightdash’s explore view and AI agents work well when the dbt YAML is well documented.

If your analytics engineer maintains every metric in dbt and the BI tool must not become a second source of truth: pick a pattern 1 or pattern 2 tool (Lightdash, Steep, or a Semantic Layer connector), or keep metrics materialized in mart tables and use a tool that reads columns without redefining them.

If you need SAML SSO and SCIM: Omni, Sigma, Tableau Cloud, Metabase Pro, Lightdash Enterprise, Steep Enterprise, Cube Enterprise, and Basedash Enterprise document both. Hex supports SSO through OIDC only.

If you want to embed dbt-modeled dashboards in your product: Lightdash (embedding add-on), Omni, Sigma, Metabase Pro, Cube Premium, and Basedash Enterprise offer signed embeds or React SDKs. For a fuller list, see our embedded analytics platform comparison.

If budget is the deciding factor: Lightdash and Metabase are free to self-host, Steep is free for up to 3 seats, and Hex and Cube have free plans. Lightdash Cloud Pro and Basedash Startup are flat monthly prices rather than per seat.

Common mistakes when picking a BI tool for dbt

• Treating “dbt integration” as one feature. Ask whether the tool reads your project, queries the Semantic Layer, or only reads descriptions. The marketing phrase is the same for all three.
• Planning on the Semantic Layer while on dbt Core. Open source MetricFlow runs locally, but the APIs that BI tools call need a paid dbt platform plan.
• Ignoring dbt version support. ThoughtSpot’s sync supports dbt 1.7 and earlier; Lightdash documents Fusion. If you are on dbt 1.8 or later or moving to dbt Core v2, check before you buy.
• Rebuilding metrics in the BI tool “just for now.” Temporary BI-side metrics become permanent and drift from dbt. Decide where each metric lives and enforce it.
• Skipping exposures. Declare dashboards as dbt exposures, or use a tool that pushes them (Omni, dbt-metabase), so you can see which reports break when a model changes.

Frequently asked questions

Which BI tools read dbt models natively?

Lightdash and Omni read the dbt project itself: Lightdash builds its semantic layer from your dbt YAML, and Omni compiles your dbt repo into a manifest and syncs changes back through pull requests. Both support dbt Core and the dbt platform. ThoughtSpot generates its models from manifest.json and catalog.json for dbt 1.7 and earlier, and Sigma imports dbt descriptions and test results. Cube imports dbt models into its own semantic layer. Metabase relies on the community dbt-metabase tool, and Basedash reads the descriptions dbt persists to the warehouse.

Which BI tools connect directly to the dbt Semantic Layer?

dbt’s integration list as of October 2026 includes Tableau, Power BI, Hex, Mode, Sigma (preview), Steep, Omni, Dot, Push.ai, Klipfolio PowerMetrics, Google Sheets, and Excel, plus custom integrations over the JDBC, ADBC, and GraphQL APIs or the Python SDK. Steep and Tableau query metrics live. Power BI’s connector is in preview and is DirectQuery only. Omni and Hex import MetricFlow definitions into their own models rather than querying the API on every request. All of these need a paid dbt platform plan.

Can I use the dbt Semantic Layer with dbt Core?

Partly. Since MetricFlow moved to the Apache 2.0 license in October 2025, dbt Core users can define metrics and query them locally from the command line. The hosted Semantic Layer APIs and the BI integrations built on them require a dbt Starter, Enterprise, or Enterprise+ plan; the free Developer plan has no API access. dbt Core teams that want governed metrics in a BI tool usually pick Lightdash, which reads MetricFlow definitions from the compiled manifest, or Omni, which imports them from the project.

We use dbt Core and Snowflake and want self-service BI for a team that doesn’t write SQL. Which tools fit?

Omni and Lightdash both read dbt Core projects on Snowflake. Omni adds a spreadsheet-style interface and an AI agent, and Lightdash works best when the dbt YAML has clear descriptions. Sigma suits spreadsheet-minded teams and accepts uploaded dbt Core artifacts in beta. ThoughtSpot imports dbt metrics on Snowflake connections and adds search and Spotter agents. Basedash fits when metrics are already materialized in Snowflake marts and the ops team wants to ask questions in plain English, though its row-level security does not cover Snowflake.

How do I stop defining metrics twice, once in dbt and once in my BI tool?

Choose one home for each metric and pick a tool that respects it. If metrics live in dbt, use a tool that reads the project (Lightdash, Omni) or queries the Semantic Layer (Steep, Tableau), or materialize metrics as columns in mart tables that any tool can read. If metrics live in the BI tool, keep dbt to transformation and document that rule. Duplication usually starts when an integration only syncs descriptions and analysts rebuild metrics in the BI layer out of habit.

Is MetricFlow open source now?

Yes. dbt Labs relicensed MetricFlow under Apache 2.0 at Coalesce in October 2025, starting with version 0.209.0, as part of its support for the Open Semantic Interchange. That spec published version 0.1 in January 2026 and entered the Apache Incubator as Apache Ossie in July 2026. Open source MetricFlow lets any tool read and compile dbt metric definitions; Lightdash already reads them from the manifest without the dbt Semantic Layer API.

Which BI tools show dbt column descriptions and test results to business users?

Sigma displays dbt table and column descriptions, dbt test results, and last run times next to the data, from the dbt platform or uploaded dbt Core artifacts. Lightdash shows dbt descriptions as hover text on every dimension and metric. Hex brings in dbt metadata through the dbt Discovery API on the dbt platform. Metabase gets descriptions through dbt-metabase. Basedash reads descriptions that dbt writes to the warehouse with persist_docs and uses them as AI context when answering questions.

Which BI tools make it easiest to reuse dbt models in embedded dashboards?

Lightdash embeds dashboards built directly on dbt models through JWT-scoped iframes or a React SDK, priced per load as an add-on. Omni offers signed iframe embeds and an Embed SDK on top of its dbt-synced model. Cube serves dbt-derived metrics to embedded dashboards and chat on Premium. Sigma and Metabase Pro embed with JWT and React SDKs, and Basedash Enterprise embeds dashboards and its AI chat with a React SDK. Row-level security for multi-tenant embeds works differently in each, so test it with your tenant column.

Related comparisons

• Best BI tools with built-in semantic layers
• Best semantic layer tools compared
• Where to define business metrics
• Basedash vs Lightdash
• Lightdash alternatives