Hex vs Sigma vs Omni (2026): pricing, AI, semantic models, and governance for growth-stage data teams
Rachel van der Lugt
Rachel van der LugtFounding Enterprise GTM at Basedash
· November 20, 2026

Rachel van der Lugt
Rachel van der LugtFounding Enterprise GTM at Basedash
· November 20, 2026

Hex, Sigma, and Omni answer different questions. Hex is the pick when your analysts live in SQL and Python notebooks and publish apps for everyone else. Sigma is the pick when business users think in spreadsheets and you want them working directly on Snowflake, BigQuery, or Databricks. Omni is the pick when you want one governed semantic model, managed in Git, behind every dashboard and AI answer. As of November 2026, only Hex publishes prices; Sigma and Omni are quote-only, and all three now charge credits for AI.
Everything below was checked against each vendor’s official pricing pages and documentation on November 20, 2026. For two-way comparisons, see Hex vs Sigma, Hex vs Omni, and Omni vs Sigma. If you are leaving Hex entirely, our Hex alternatives guide covers a wider set.
| Attribute | Hex | Sigma | Omni |
|---|---|---|---|
| Core interface | SQL and Python notebooks, published data apps, Threads chat | Spreadsheet-style workbooks on live warehouse data | Point-and-click workbooks, SQL, and spreadsheet tabs over a shared semantic model |
| Published price | Community free; Professional $36 and Team $75 per editor per month; Enterprise custom | No, quote only (View, Act, Analyze, Build licenses) | No, quote only |
| AI pricing | Monthly credit grant per seat; 50-credit top-up pack $25 | Agent messages 0.1 credit each since November 1, 2026; credit price via sales | Included allotment; overage $1.00 per credit |
| Free trial | Team plan, 14 days, no card | 7 days | Request form; length not published |
| Deployment | Cloud (US or EU multi-tenant); single tenant add-on on Enterprise | Cloud on AWS, Azure, or GCP; region fixed at creation | Cloud on AWS or Azure in eight regions |
| Query model | Live queries, results cached 60 minutes; DuckDB for in-memory dataframe SQL | Live queries; Sigma “does not store your data” | Live queries, results cached 6 hours by default |
| Semantic layer | Semantic models on Team and Enterprise; sync from dbt MetricFlow, Cube, Snowflake semantic views (beta) | Data models with metrics and relationships, manageable as JSON or YAML; dbt Semantic Layer (dbt Cloud only) | Shared YAML model with branches and Git pull requests; bi-directional dbt integration |
| Row-level security | Warehouse permissions through OAuth connections (Enterprise; Snowflake, Databricks, BigQuery) | User-attribute formulas in data models; column-level security | Access filters on model topics; field-level access grants |
| SSO and SCIM | OIDC only, no SAML (Enterprise); directory sync (Enterprise) | SAML 2.0 or OIDC; SCIM (requires SAML) | SAML or OIDC; SCIM for Okta, Entra, Rippling |
| Natural language for business users | Threads, needs an Explorer seat or higher | Sigma Assistant and Sigma agents (GA October 2026) | Omni Agent, on by default |
| MCP server | Yes, Team and Enterprise | Yes | Yes |
| Embedding | Signed embedding with an add-on (Enterprise) | Public and secure embeds; secure is a premium feature; React SDK | Signed iframe URLs; @omni-co/embed server SDK |
| Best for | Analyst teams that write code and publish apps | Spreadsheet-native business users on a cloud warehouse | Data teams that want governed self-serve and AI on one model |
| Not ideal for | Large non-technical audiences on a budget; SAML-only IdPs | Teams that need published prices or Git-native modeling | Teams without anyone to own the semantic model |
Pricing verified November 20, 2026.
The criteria come from the questions growth-stage teams ask when these three share a shortlist:
Hex is the only one of the three with list prices. Community is free, Professional is $36 per editor per month, Team is $75 per editor per month, and Enterprise is custom (Hex pricing). Every plan allows unlimited users, but seat type matters. Editors build notebooks and apps; viewers open apps and shared Threads, and viewer seats “may be free or paid depending on your pricing plan” (Hex roles). Asking questions in Threads or using Explore needs an Explorer role or higher, a per-seat add-on on Team and Enterprise priced through sales. AI runs on credits: 30 a month per Professional editor, 40 per Team editor, 10 per Team Explorer, with 50-credit top-ups for $25 (Hex credits).
For a 25-person team with five editors on Team, the published part is $375 a month; the rest depends on how many of the other 20 need Explorer seats.
Sigma doesn’t publish prices; its pricing URL redirects to a contact form, and the license overview tells buyers to contact their account executive. Licenses are tiered by what people do: View, Act (forms and input tables), Analyze (ad hoc analysis), and Build (create and share workbooks). Since November 1, 2026, each message sent to a Sigma agent costs 0.1 credit, and the credit price is also quote-only (billable usage events).
Omni doesn’t publish prices either; there is no public pricing page, and trials start from a request form. Each account gets an included AI allotment, pooled across the account, and usage beyond it is “billed at $1.00 per credit” once an admin enables overage (Omni AI usage).
None of the three is purely usage-based: Hex and Sigma charge by seat type, Omni by negotiated contract, and all three meter AI on top. Our BI pricing breakdown compares seat limits and AI costs across more tools.
Sigma is the most familiar for spreadsheet users. Workbooks look and behave like a spreadsheet over live warehouse tables, input tables let people add their own rows, and Sigma Assistant is a “natural language query (NLQ) interface” for asking questions in plain English (Sigma Assistant). In October 2026, Assistant in the workbook became generally available for building and editing dashboards too.
Omni gives three ways in: natural language, a visual point-and-click interface, or SQL (Omni querying). Spreadsheet tabs offer Excel-style calculations on live data. Because every query runs through the shared model, business users pick from curated fields instead of raw tables.
Hex is built for people who write code. Non-technical users mostly consume published apps, and Hex’s self-serve tools, Threads for chat and Explore for drag-and-drop drill-down, need an Explorer seat. Explore from an app or a data object is a Team and Enterprise feature (Hex Explore).
Setup effort follows the same pattern. All three query the warehouse live, so none needs an extract pipeline. Sigma needs the least before business users get value: connect the warehouse and they can start on lightly modeled tables. Omni starts from a schema model that mirrors your database, but self-serve and AI quality depend on someone curating topics and fields. Hex is fastest for an analyst who writes SQL and slowest for everyone else, who need published apps or semantic models first.
For a team without a dedicated data engineer, Sigma and Omni are easier for business users than Hex. Sigma wins for people who already think in spreadsheets; Omni wins once someone has curated the model.
Omni puts the semantic layer at the center: a schema model that mirrors the database, a shared governed model, and workbook models, where “changes you make in a workbook land in the workbook model first” until someone promotes them (model promotion). The model lives in YAML, can require Git pull requests (Omni Git), and the dbt integration is bi-directional: it reads dbt metadata and the dbt Semantic Layer and pushes descriptions and exposures back (Omni dbt). The Modeling Agent, announced in April 2026, drafts views, relationships, dimensions, and measures from your tables.
Sigma uses data models with reusable metrics and defined relationships. Since April 2026, data models can be managed as JSON or YAML through an API, and in November 2026 workbooks and reports followed (data models as code). Sigma reads the dbt Semantic Layer, but “the dbt Semantic Layer is not available for dbt Core integrations,” so you need dbt Cloud (dbt Semantic Layer in Sigma). Sigma doesn’t document a native Git integration; the as-code API is the path to CI/CD.
Hex added semantic models on Team and Enterprise. You can author them in Hex’s Modeling Workbench or import them with Semantic Model Sync, which is in beta and supports dbt MetricFlow, Cube, and Snowflake semantic views (semantic model sync). The dbt path imports MetricFlow YAML rather than querying the live dbt Semantic Layer. Git support is a one-way export of projects to GitHub, GitLab, or Bitbucket on Team and Enterprise (Hex Git export).
For a Series B team on Snowflake that wants semantic modeling plus collaboration, Omni has the most complete model and Sigma the closest thing to spreadsheet collaboration. Our comparison of BI tools with built-in semantic layers covers this in more depth.
Omni enforces row-level security in the model. Access filters limit “access to rows in a dataset based on user attributes,” and access grants restrict individual fields (access filters). It supports SAML and OIDC, SCIM through Okta, Microsoft Entra, and Rippling, and audit logs delivered to S3, Google Cloud Storage, or Azure Blob Storage (audit logs).
Sigma enforces row-level security with formulas such as CurrentUserAttributeText("Region") = [Store Region] (row-level security) and offers column-level masking in data models. It supports SAML 2.0 and OIDC. SCIM requires the organization to use SAML (Sigma SCIM). Audit logging is off by default and must be enabled by Sigma support.
Hex has the narrowest set. Single sign-on is “via OpenID Connect (OIDC). SAML is not supported,” and it is an Enterprise feature (Hex SSO). Directory sync for Okta, Rippling, Google, and Azure, audit logs, and OAuth data connections are Enterprise too. Row-level security for workspace users comes from the warehouse: OAuth connections “run each query under a user’s warehouse identity,” and they work with Snowflake, Databricks, and BigQuery (OAuth connections). Hex-native row filtering is documented for signed embedding, not for internal users.
For a fintech or other regulated team on BigQuery that needs strong governance, Omni and Sigma both cover SAML, SCIM, and in-tool row-level security. Hex works if your identity provider supports OIDC and you are comfortable enforcing rows in BigQuery itself.
Omni Agent answers questions, generates queries and charts, and builds dashboards (Omni AI). It is “enabled by default,” runs on Amazon Bedrock or Microsoft Foundry depending on your deployment, and accepts bring-your-own models from Anthropic, OpenAI, Vertex AI, Snowflake Cortex, and others (LLM settings). Omni also has a Workbook Agent and an MCP server that can scope queries to a user’s permissions.
Sigma made Sigma agents and Assistant in the workbook generally available in October 2026 (Sigma release notes). AI runs either through a warehouse (Snowflake, BigQuery, Databricks) or an external provider such as OpenAI, Anthropic, Gemini, or Bedrock, and Sigma’s docs note “you cannot choose which model Sigma uses” (supported AI models). Sigma’s MCP server lets Claude, ChatGPT, Cursor, and others search, query, and build workbooks.
Hex has the Hex agent and Notebook agent for analysts, Threads for conversational questions, Context Studio for curating what the agent knows, and a governed mode that limits the agent to endorsed sources. Hex’s own docs say governed mode “isn’t a hard security boundary” (governed mode). Its MCP server is on Team and Enterprise, and Hex 4.0, in public beta since October 2026, backs projects with code files.
For AI answers to business users without a full-time data engineer, Omni’s model-grounded agent and Sigma’s Assistant need less setup than Hex’s Threads, which also needs Explorer seats.
Hex is the better notebook: multiplayer editing with cell locking (real-time collaboration), SQL and Python in one project, and in-memory DuckDB for querying dataframes. Omni is the better governed dashboard layer, because every dashboard and agent answer inherits the same model and access filters. Some teams run both, Hex for the analytics team’s deep work and Omni for company-wide dashboards. If you can only buy one, choose by who your main users are.
If your team is small, has no data engineer, and wants AI to do most of the building, Basedash is a different shape worth comparing: people ask questions in chat and the AI data analyst writes the SQL, builds charts, and assembles dashboards.
Basedash fact card (verified November 20, 2026)
basedash.groups session variable (row-level security). Schema, table, and column access rules hide data from the AI on any source.@basedash/embed React SDK.Hex is the only one with published prices: Professional costs $36 and Team $75 per editor per month as of November 2026, so a few editors and many viewers can be inexpensive. Costs rise when business users need Explorer seats to ask questions in Threads or use Explore, since Explorer is a sales-priced add-on. Sigma and Omni are quote-only, so a direct comparison needs real quotes. All three meter AI separately: Hex sells 50 credits for $25, Sigma charges 0.1 credit per agent message, and Omni bills $1.00 per credit over its included allotment.
Sigma is better for non-technical users. Its workbooks work like a spreadsheet on live warehouse data, and Sigma Assistant answers natural language questions. Hex is built around SQL and Python notebooks; non-technical users mostly view published apps, and asking questions in Threads or drilling down with Explore requires an Explorer seat, which is an add-on on Team and Enterprise. Hex suits teams where a few analysts build and everyone else consumes. Sigma suits teams where business users want to build their own analysis.
No. As of November 2026, Hex’s documentation says it supports single sign-on through OpenID Connect (OIDC) and that SAML is not supported. SSO is on the Enterprise plan, along with directory sync for Okta, Rippling, Google, and Azure. Most major identity providers support OIDC, so this is rarely a blocker, but check with your IT team if your security policy requires SAML. Sigma and Omni both support SAML and OIDC.
Omni’s semantic layer is more central and more developer-friendly. Every query runs through a shared YAML model that can be managed in Git with pull requests, and its dbt integration both reads dbt metadata and the dbt Semantic Layer and writes back. Sigma’s data models define metrics and relationships and can be managed as JSON or YAML through an API, but Git is not native, and its dbt Semantic Layer support requires dbt Cloud. Pick Omni if consistent metric definitions matter most; pick Sigma if spreadsheet flexibility matters more.
Yes. All three connect live to Snowflake, BigQuery, Databricks, Redshift, and Postgres, along with others such as ClickHouse and MotherDuck (both in beta on Sigma). The differences are in governance. Hex can run queries under each user’s own Snowflake, BigQuery, or Databricks identity through OAuth connections on Enterprise. Sigma can assign warehouse roles from user attributes and enforce row-level security with formulas. Omni applies access filters from its model to every query.
Choose by who the main users are. If the data team is two to five analysts who write SQL and Python and publish results, Hex is the better fit and its pricing is published. If the goal is to let the rest of the company self-serve from consistent definitions, Omni is the better fit, because its shared model and access filters apply to every dashboard and AI answer. Some startups begin with Hex for analysis and add Omni or Sigma once business users need their own dashboards.
Written by

Founding Enterprise GTM at Basedash
Rachel van der Lugt is founding enterprise GTM at Basedash, where she leads enterprise go-to-market for an AI-native business intelligence platform. Her work focuses on helping larger teams evaluate, adopt, and roll out Basedash for governed analytics across the organization.
Basedash lets you build charts, dashboards, and reports in seconds using all your data.