Short answer
For a small team on a budget in 2026, Helical Insight is the best overall pick: Community Edition is free, includes every feature, and mixes SQL/code workflows with no-code for business users. Evidence Core and self-hosted Lightdash are the strongest pure BI-as-code options. Apache Superset, Metabase and Redash are the leading free SQL-first alternatives.
TL;DR: Which code-centric BI tool should a small team start with?
- Start with Helical Insight if the team mixes analysts, developers and business users, and you need dashboards, pixel-perfect reports, embedding, SSO and AI without per-user fees.
- Choose Evidence Core or Lightdash OSS if Git, SQL and a metrics-in-code workflow are the main requirement, and you can self-host. Their hosted plans start at $2,500 and $3,000 a month, which is rarely “on a budget.”
- Choose Apache Superset, Metabase or Redash if you want a free, SQL-first stack and can accept operational effort (and, for Metabase, paid upgrades for SSO, white-labeling and signed embeds).
- Treat Rill, Observable Framework and Grafana as specialists: operational metrics-as-code, static data apps, and observability dashboards, not full business-BI suites.
Key facts at a glance
- Helical Insight Community Edition is free under AGPL-3.0 and includes every product feature, with no per-user fee. Enterprise is a flat, unpublished support license (Community Edition, GitHub).
- Evidence Core is MIT-licensed and free to self-host. Evidence Studio Team is $2,500/month with unlimited users; SSO, row-level rules, embedding and white-labeling are listed on Enterprise (Evidence pricing, self-host docs).
- Lightdash is MIT-licensed and free to self-host. Cloud Pro starts at $3,000/month with no per-seat fee. Embedding and AI agents are not in the open-source edition (Lightdash pricing).
- Apache Superset is Apache 2.0 and free to self-host. The latest release we found is 6.1.0 (13 May 2026). Preset, the managed service, is $0 for up to 5 users, then Professional at $20/user/month billed annually ($25 billed monthly). Embedded viewer licenses start at $500/month for 50 viewers (Preset pricing).
- Metabase open source is AGPL and free, with unlimited users. Cloud Starter is $100/month for 5 users ($6 per extra user; $1,080/year plus $65 per extra user annually). Pro is $575/month for 10 users ($12 per extra user) and is the plan for SSO, row/column security, white-labeling and signed embeds. Prices on Metabase’s reference were last marked 9 November 2026 (Metabase pricing).
- Redash is BSD-2-Clause and free to self-host. The official hosted service closed in 2021. The project is community-maintained; the latest release is v26.9.0 (24 September 2026).
- Rill Developer is Apache 2.0 for local, Git-based project work. Official docs say Developer is not a standalone replacement for shared production. Rill Cloud Starter is $20/seat/month (up to 20 seats and 3 projects) plus compute at $0.15 per compute-unit-hour; plans include a $250 usage credit. Embedding on Starter is previews only (Rill pricing).
- Observable Framework is ISC-licensed. The latest release is v1.13.4 (2 March 2026). Observable Cloud was deprecated in April 2025; you self-host the static site (GitHub).
- Grafana open source is AGPL-3.0 and free to self-host. Grafana Cloud has a free tier and Pro from $19/month plus usage. The Cloud visualization free tier is limited to 3 active users.
What is a code-centric BI tool?
A code-centric BI tool is a business intelligence product that treats analytics as software: SQL or markdown in a repository, metrics defined in code, and changes that can go through Git, reviews and CI.
People also call this BI-as-code or SQL-first BI. The idea is the same. Instead of only clicking together a dashboard that lives inside a vendor’s app, the team can:
- write SQL (or a documented semantic layer) as the source of truth for metrics
- keep models, metrics and sometimes dashboards in Git, with diffs and pull requests
- deploy or regenerate reports from a CLI, API or pipeline
- let viewers open a browser UI without having to touch the code
It is not the same thing as a notebook, a Python app, or a semantic-layer-only product. Those can be part of a code-centric stack, but a BI tool still has to serve dashboards, access control and, ideally, a way for non-authors to consume the result.
Helical Insight sits slightly to one side of the purest BI-as-code tools. It is a full BI platform with a browser designer, APIs and SQL, not a Markdown-first framework. We still include it first because a small team on a budget usually needs both a code-friendly core and a no-code path for everyone else.
How should a small team on a budget choose a code-centric BI tool?
Score each tool on the cost of 5, 10 and 25 users, whether you can self-host, how steep the learning curve is, whether Git is a first-class workflow, and whether non-technical people can view or build content.
For this ranking we used six practical tests:
- Cost at 5, 10 and 25 users. We used only published 2026 prices. “$0” means no license fee; you still pay for servers and people’s time.
- Self-hosting. Can you run it on your own Docker host or VM without a mandatory cloud contract?
- Learning curve. Can a two- or three-person data team ship something in days, not months?
- Git and code workflow. Are metrics, SQL or dashboards reviewable as text?
- Non-technical viewer access. Can the rest of the company use the output without a seat cliff or a Git account?
- Embedding, SSO and maintenance. What is included in the free edition, and how much operational work does the tool create?
Hosted “unlimited user” plans that start at $2,500 or $3,000 a month failed the budget test for most small teams, even when the product itself is excellent. In those cases we ranked the open-source or self-hosted edition, and we say so plainly.
What are the best code-centric BI tools for a small team on a budget?
The best overall tool for a budget-conscious small team in 2026 is Helical Insight, followed by Evidence Core and self-hosted Lightdash for pure BI-as-code, then Apache Superset, Metabase and Redash for free SQL-first stacks. The nine tools below are ranked for that specific job, not for enterprise observability or custom data-app development.
1. Helical Insight
Helical Insight is the best code-centric BI tool for a small team on a budget because the free Community Edition already includes the enterprise features other tools sell as upgrades.
- What it is: An open-source, browser-based BI platform from Helical IT Solutions, with a Java backend, a React frontend and JDBC connections. Analysts can work in SQL and APIs; business users can build in a drag-and-drop designer.
- Code model: SQL, a governed metadata layer, REST APIs and an extensible framework. It is not Markdown-as-code in the Evidence sense, but definitions, security and embeds can be managed in a repeatable, developer-friendly way.
- License: Community Edition is AGPL-3.0 and free, with every feature included. Enterprise is a commercial, flat-priced support license that removes Helical branding.
- Price for a small team: $0 on Community Edition. There is no public Enterprise list price and no per-user fee.
- Best for: Mixed teams that need self-service dashboards, pixel-perfect canned reports, embedding and multi-tenancy, white-labeling, SSO and Instant BI chat with a bring-your-own LLM, including local models. Docker is documented for self-hosting.
Start with the free Community Edition
Dashboards, pixel-perfect reports, embedding, SSO and AI — no per-user fees.
2. Evidence
Evidence is the strongest pure BI-as-code tool on this list, as long as you use Evidence Core and can live with a narrower self-hosted feature set.
- What it is: A SQL-and-Markdown reporting framework at evidence.dev. Evidence Core is the MIT-licensed open-source framework. Evidence Studio is the hosted commercial platform. Some pages still use the older name Evidence Cloud.
- Code model: Reports are Markdown (Markdoc) with SQL and components, stored in Git. Core talks to one warehouse at a time, in that warehouse’s native dialect. YAML metrics are reusable. This is the textbook BI-as-code workflow. See our full Helical Insight vs Evidence comparison.
- License: Core is MIT. Studio is a commercial subscription.
- Price for a small team: Core is $0 to self-host. Studio Team is $2,500/month for unlimited users, with 20,000 AI credits included and extra credits at $0.01. Enterprise is a custom annual contract.
- Best for: Data teams that already want pull requests on reports, static validation, and an agent-friendly text format.
- Limitations: Official self-host docs include only HTTP Basic Auth out of the box. Evidence Warehouse, SQL models, SSO, row-level security, the web editor, agents, scheduled reports and embedding are Studio features. Embedding and white-labeling are listed on Enterprise, not Team. For a ten-person company, $2,500 a month is rarely a budget option.
3. Lightdash
Lightdash is the best code-centric BI tool if your metrics already live in dbt or in YAML next to the warehouse.
- What it is: An open-source BI tool built around a semantic layer, a metrics explorer and “charts as code.” dbt is optional; you can point standalone Lightdash YAML at a warehouse.
- Code model: Metrics and, on paid plans, charts and dashboards can be managed as code and synced with GitHub, GitLab, Azure DevOps or Bitbucket.
- License: The self-hosted core is MIT. Cloud Pro and Enterprise are commercial.
- Price for a small team: Self-hosted open source is $0, with unlimited users and community Slack support. Cloud Pro starts at $3,000/month with unlimited users and a 21-day trial. Enterprise is a quote. AI agents, MCP and embedding are not in the open-source edition; Pro embedding is usage-based (first 1,000 loads free, then $0.05/load, or $790/month for 100,000 loads per embed worker).
- Best for: dbt-native teams that want governed self-service on top of version-controlled metrics. See Helical Insight vs Lightdash.
- Limitations: $3,000 a month is not a small-team budget. On the free edition, SSO is Google-only, and embedding, AI, APIs, user groups and custom roles are absent. Self-hosting is a real DevOps job (Docker or Kubernetes).
4. Apache Superset
Apache Superset is the best free, full-featured SQL visualization platform if your team can operate it.
- What it is: An Apache Software Foundation BI project with SQL Lab, a chart builder, dashboards and a semantic layer. The latest release we found is 6.1.0, dated 13 May 2026. Preset is a separate company that sells managed Superset.
- Code model: SQL-first datasets and charts, plus configuration in Python/YAML. Teams often store dashboard export assets in Git, though it is less “reports as Markdown” than Evidence.
- License: Apache License 2.0, including commercial use, with no paid edition of the open-source project itself.
- Price for a small team: Self-host $0 for any number of users (you pay infrastructure and maintenance). Preset Starter is $0 for up to 5 users. For 10 users you leave the free Starter plan: Professional is $200/month billed annually ($20 × 10) or $250/month billed monthly. For 25 users that is $500/month annually or $625 billed monthly.
- Best for: Technical teams that want a powerful, widely adopted SQL-first BI server and can handle upgrades, a metadata database and Redis. See Helical Insight vs Apache Superset.
- Limitations: Operation is heavier than a single Docker BI app. Pixel-perfect canned reports are not a core strength. Preset embedding is an add-on from $500/month for 50 viewer licenses; self-hosted Superset does not add that fee. SSO via OAuth or LDAP can be configured in the open-source project.
5. Metabase
Metabase is the friendliest free BI tool for non-technical users, and it can be code-centric, but the most code- and enterprise-friendly features sit on Pro.
- What it is: A popular open-source BI product with a visual query builder, a SQL editor, dashboards and subscriptions.
- Code model: Authors can write SQL. Serialization — export and import content as YAML — is a Pro feature, so a true Git workflow for dashboards is not in the free edition.
- License: Open source is AGPL. Cloud and self-hosted commercial editions use Metabase’s commercial terms.
- Price for a small team: Open source is $0 with unlimited users. Cloud Starter is $100/month for 5 users ($1,080/year); extra users are $6/month or $65/year, so 10 users cost $130/month ($1,405/year). Pro is $575/month for 10 users ($6,210/year), or $12/month per extra user, so 25 users cost $755/month. Pro can be cloud or self-hosted at the same published price. Enterprise starts at $20,000/year.
- Best for: Small teams that want the shortest path to self-service questions, and are happy to self-host the AGPL edition.
- Limitations: Guest embeds exist on the open-source edition and show a “Powered by Metabase” banner. SSO (SAML, JWT or OIDC), row and column security, white-labeling, SSO embeds and the React SDK require Pro. That is a steep jump from $0 or $130/month to $575/month if those features matter.
6. Redash
Redash is the leanest SQL-first dashboard tool, and it is free — with the honest caveat that you are adopting a community-maintained project.
- What it is: A browser-based query, visualization and dashboard tool for SQL and NoSQL sources. Databricks acquired Redash in 2020 and closed the hosted service on 30 November 2021. Volunteers have kept releasing the open-source project; v26.9.0 arrived on 24 September 2026.
- Code model: Saved SQL queries are the core object, with a REST API for everything in the UI. This is classic SQL-first BI, not a semantic-layer-as-code tool.
- License: BSD-2-Clause.
- Price for a small team: $0 to self-host, at any team size. There is no official paid cloud today.
- Best for: Engineers who want a query editor, scheduled refreshes, alerts and shareable dashboards without a vendor contract. See Helical Insight vs Redash.
- Limitations: You operate PostgreSQL, Redis and the app yourself. There is no vendor SLA. Pixel-perfect reporting, a first-class semantic layer and a polished no-code builder are not the point. Open-source authentication includes Google OAuth, SAML, LDAP and JWT, which is better than many free tiers.
7. Rill
Rill is a strong Git-based, SQL-and-YAML BI tool for fast operational dashboards, but shared production use is a paid cloud product.
- What it is: A BI-as-code workflow split into Rill Developer (local IDE-like environment) and Rill Cloud (shared production). Projects are SQL and YAML in Git, often used with an OLAP engine for interactive “slice and dice” dashboards.
- Code model: Sources, SQL models and metrics views are files. Official docs are explicit that Developer and Cloud are meant to be used together, and that Developer is not a sufficient replacement for Cloud.
- License: The GitHub project is Apache 2.0. Rill Cloud is commercial.
- Price for a small team: Local Developer use is $0. Cloud Starter is $20/seat/month, up to 20 seats and 3 projects, plus compute at $0.15 per compute-unit-hour and storage fees if you use Rill-managed infrastructure. All Cloud plans include a $250 usage credit and a trial. Ten seats are $200/month before compute. Growth is $30/seat/month. Enterprise is custom. Starter embedding is previews only; Growth embedding is $500/month for up to 100 active embedded seats.
- Best for: Teams that want metrics-as-code and very fast exploratory dashboards, and can budget Cloud for sharing.
- Limitations: A genuine small-team budget path is weaker here than for Helical, Evidence Core or Superset, because official shared production is Cloud-priced and usage-based. Do not assume Developer is a free self-hosted Cloud.
8. Observable Framework
Observable Framework is an excellent open-source way to build data apps as code, but it is not a multi-user BI platform.
- What it is: A static-site framework for data apps, dashboards and reports, written in Markdown, JavaScript and data loaders. The latest release is v1.13.4 (2 March 2026). Observable Cloud was deprecated in April 2025, and the
deploycommand later gained a deprecation notice. - Code model: The entire site is code in Git. You build a static bundle and host it on any static host, or in Docker behind your own auth.
- License: ISC, which is a permissive open-source license.
- Price for a small team: $0 for the framework. Hosting is whatever you already pay for a static site or container.
- Best for: Developers who want custom, explainable data products and are willing to build navigation, access control and scheduling themselves.
- Limitations: There is no built-in semantic layer, pixel-perfect bursting, organization SSO catalog, or vendor-supported embed API. Framework remains open source, but Observable’s commercial focus has moved toward notebooks and canvases, so you should treat this as a library you own, not a BI product you buy support for.
9. Grafana
Grafana is the right code-centric choice when the “BI” you need is operational dashboards as code, not business reporting.
- What it is: A widely used visualization and dashboard platform, strongest for metrics, logs and traces, with SQL data sources as well. Grafana documents dashboard-as-code workflows, including Git Sync, file provisioning, Terraform and a Foundation SDK.
- Code model: Dashboards as JSON or generated code, provisioned from Git.
- License: Grafana open source is AGPL-3.0. Grafana Cloud and Enterprise are commercial.
- Price for a small team: Self-hosted open source is $0. Grafana Cloud has a free tier and Pro from $19/month plus usage. The Cloud visualization free tier is 3 active users, so it is not a 10-user business-BI license. Enterprise-oriented Cloud spend starts at a $25,000/year commit on the full-service tier. Some data-source plugins and SAML-style SSO sit in commercial editions.
- Best for: Engineering and SRE teams that already live in Grafana and want the same as-code discipline for operational charts.
- Limitations: It is not a substitute for pixel-perfect financial packs, a business semantic layer, or mixed business-user self-service of the Metabase or Helical kind.
How do these tools compare side by side?
If you only look at published license cost for ten users, several tools can be $0. The differences are what the free edition actually includes.
| Tool | License | Self-host | Cost for 10 users (published) | Code model | Non-tech friendly | Embedding on free/self-host | SSO on free tier |
|---|---|---|---|---|---|---|---|
| Helical Insight | AGPL-3.0 (CE) | Yes | $0 | SQL, APIs, metadata | Yes, drag-and-drop plus SQL | Yes, all editions | Yes (JWT, OAuth, Okta, Keycloak) |
| Evidence | MIT (Core) | Core: yes | Core $0; Studio Team $2,500/mo unlimited | SQL + Markdown in Git | Viewers yes; authors need code | No (Studio Enterprise) | Core: HTTP Basic only |
| Lightdash | MIT (OSS) | Yes | OSS $0; Cloud Pro from $3,000/mo unlimited | dbt / YAML metrics as code | Explorer on prepared metrics | No (paid add-on on Pro) | Google SSO only on OSS |
| Apache Superset | Apache 2.0 | Yes | Self-host $0; Preset Pro $200/mo annual or $250 monthly | SQL Lab + config as code | Mixed; SQL often required | Yes on self-host; Preset add-on from $500/mo | OAuth/LDAP can be configured |
| Metabase | AGPL (OSS) | Yes | OSS $0; Cloud Starter $130/mo; Pro $575/mo | SQL + query builder; YAML serialization on Pro | Yes | Guest embeds with banner | No (SAML/JWT/OIDC are Pro) |
| Redash | BSD-2-Clause | Yes | $0 | Saved SQL + API | Viewers yes; authors need SQL | Share/embed links in OSS | Google OAuth, SAML, LDAP, JWT in OSS |
| Rill | Apache 2.0 (Developer) | Local Developer; sharing via Cloud | Cloud Starter $200/mo + compute | SQL + YAML in Git | Cloud dashboards for consumers | Previews only on Starter | Google/Microsoft on Cloud (paid) |
| Observable Framework | ISC | Yes (static site) | $0 | JS/Markdown data apps as code | No | You host and embed the site | You add your own |
| Grafana | AGPL-3.0 (OSS) | Yes | OSS $0; Cloud Pro from $19/mo + usage | Dashboards as code | Mixed; ops-oriented | Supported in OSS, with commercial extras | OAuth/LDAP in OSS; SAML typically commercial |
Prices are published subscription or license fees, checked 20 November 2026, and exclude your own servers, warehouse query costs and taxes. Cloud plans change; confirm on the vendor site before you buy.
Which tool should you choose?
Match the tool to the team you actually have, not to the most fashionable BI-as-code blog post.
You are a two-analyst startup with a founder who still wants to click a dashboard. Choose Helical Insight. Community Edition covers SQL users and non-technical users in one free platform, including reports and SSO.
Your data team already reviews dbt models in Git. Choose self-hosted Lightdash if you can operate it, or Helical Insight if you also need pixel-perfect reports and embedding without a $3,000 Cloud Pro bill. Evidence Core is the alternative if reports themselves should be Markdown files.
You want reports as pull requests, and nobody needs a drag-and-drop builder. Choose Evidence Core. Stay off Studio until you need hosted agents, access control or embedding — and budget for Team or Enterprise when you do.
You have a strong SQL culture and one engineer who can run Docker, Postgres and Redis. Apache Superset or Redash will work. Prefer Superset for a richer visualization and semantic-layer story; prefer Redash for a thinner query-and-dashboard tool. Read the Redash maintenance history with your eyes open.
Non-technical teammates must build their own questions this week. Metabase open source is the fastest of the free SQL-plus-GUI tools. Plan for Pro if you will need SSO, row-level security or signed embeds.
You are shipping operational or product metrics dashboards from Git, and consumers will live in a shared cloud. Rill Cloud can be a good fit, but run the seat-plus-compute math; ten seats start at $200/month before usage.
You are building a custom narrative data site, not a BI workspace. Observable Framework is the code-centric choice. Streamlit, Dash or Quarto may be better if the team is already in Python or R — see the “also considered” section below.
Your world is already Grafana. Keep Grafana for ops. Do not force finance reporting into it.
If you are also comparing these tools with commercial suites, see Helical Insight vs Power BI, Helical Insight vs Tableau, Helical Insight vs Looker, Helical Insight vs Qlik Sense and Helical Insight vs JasperReports.
Not sure which stack fits a 10-person team?
We’ll map your warehouse, Git workflow and reporting needs to a free Community Edition proof of concept.
What does a 5-, 10- or 25-person team actually pay?
Only a few of these tools publish a simple per-user cloud price. The rest are either $0 to self-host or a flat hosted plan that does not get cheaper for a small team.
The figures below are official list prices, in US dollars, excluding tax, warehouse compute and the servers you run yourself.
| Team size | Helical Insight CE | Evidence | Lightdash | Apache Superset / Preset | Metabase | Rill Cloud |
|---|---|---|---|---|---|---|
| 5 users | $0 | Core $0; Studio $2,500/mo | OSS $0; Cloud Pro $3,000/mo | Self-host $0; Preset Starter $0 | OSS $0; Cloud Starter $100/mo ($1,080/year) | Starter $100/mo + compute, minus $250 credit |
| 10 users | $0 | Core $0; Studio $2,500/mo | OSS $0; Cloud Pro $3,000/mo | Self-host $0; Preset Pro $200/mo annual or $250 monthly | OSS $0; Starter $130/mo ($1,405/year); Pro $575/mo ($6,210/year) | Starter $200/mo + compute, minus $250 credit |
| 25 users | $0 | Core $0; Studio $2,500/mo | OSS $0; Cloud Pro $3,000/mo | Self-host $0; Preset Pro $500/mo annual or $625 monthly | OSS $0; Starter $220/mo; Pro $755/mo | Starter allows only 20 seats; Growth is $750/mo + compute |
Redash, Observable Framework and self-hosted Grafana OSS are $0 at all three sizes for the software itself. Grafana Cloud visualization’s free tier stops at 3 active users, so we do not invent a 10- or 25-user Cloud total. Rill compute is usage-based; the table shows seat fees only, plus the published $250 credit.
The budget lesson is straightforward. A small team that self-hosts Helical Insight, Evidence Core, Lightdash OSS, Superset, Metabase OSS, Redash or Grafana OSS can keep software cost at zero. The expensive line items appear when you want a vendor to host the product, or when you need SSO, embedding and white-labeling that a vendor reserved for a paid edition — unless you picked Helical Insight, which includes those in Community Edition.
Which tools did we consider but not rank?
Streamlit, Plotly Dash, Quarto and Cube are useful in a code-centric stack, but they are not complete BI platforms for a small mixed team.
- Streamlit is an Apache 2.0 Python app framework, with a free Community Cloud option (see Streamlit’s current docs for host quotas). It is excellent for internal data apps and weak as a governed BI workspace.
- Plotly Dash is a Python (and related) app framework for custom analytical UIs, not a SQL-and-semantic-layer BI server.
- Quarto is a publishing system for computational documents. It is code-centric, but it is not a shared BI catalog with SSO and scheduled bursting.
- Cube is a semantic layer that can sit under a BI tool. It is worth evaluating as a complement, not as the BI tool itself.
Frequently Asked Questions
What are the best code-centric BI tools for a small team on a budget?
Helical Insight is the best overall pick in 2026 because Community Edition is free and already includes dashboards, pixel-perfect reports, embedding, SSO and AI. Evidence Core and self-hosted Lightdash are the best pure BI-as-code options. Apache Superset, Metabase and Redash are the leading free SQL-first alternatives.
What does “code-centric BI” or “BI-as-code” mean?
It means analytics is treated like software: SQL or markdown in Git, metrics defined in code, and changes that can be reviewed and deployed. Viewers still use a browser. It is broader than “the tool has a SQL window,” and narrower than “any Python chart library.”
Is Helical Insight really free for a 10- or 25-person team?
Yes, on Community Edition. The AGPL-3.0 edition includes every product feature and has no per-user fee. You still pay for the machine you run it on. Enterprise is optional, flat-priced support; Helical does not publish that price.
Why not just pay for Evidence Studio or Lightdash Cloud?
Both are strong hosted products. They are not budget products. Evidence Studio Team is $2,500 a month and Lightdash Cloud Pro starts at $3,000 a month. If you need only the open-source Core or OSS editions, those remain $0 — with the feature limits described above.
Which free tool is best if nobody on the team wants to write Markdown?
Helical Insight or Metabase. Helical Insight adds pixel-perfect reports, embedding and SSO in the free edition. Metabase is very easy for ad hoc questions, but SSO, white-labeling and signed embeds require Pro at $575/month for 10 users.
Can a small team self-host all of these tools?
You can self-host Helical Insight, Evidence Core, Lightdash OSS, Apache Superset, Metabase OSS, Redash, Observable Framework and Grafana OSS. Rill Developer runs locally; official docs say shared production belongs on Rill Cloud. Observable Cloud is deprecated, so Framework sites are self-hosted.
Is Redash still maintained?
Yes, as a community project. Databricks closed the hosted service in 2021. The open-source repository is still releasing; v26.9.0 is dated 24 September 2026. There is no vendor-backed cloud or published paid plan.
Do I still need a semantic layer if I pick Helical Insight?
A semantic or metadata layer is still a good idea. Helical Insight includes a metadata layer in every edition, so you do not have to buy a separate product to govern metrics, though some teams still pair a warehouse modeling tool such as dbt or Cube with it.
Final Verdict: What are the best code-centric BI tools for a small team on a budget?
For most small teams in 2026, Helical Insight is the best place to start, because the free edition already behaves like an enterprise BI platform. Evidence Core and Lightdash OSS are the tools to pick when the team is truly optimizing for Git-native metrics and reports. Apache Superset, Metabase and Redash remain excellent $0 SQL-first options if you are honest about operations and about which features are paid upgrades. Rill, Observable Framework and Grafana are specialists: use them when their narrower job is the job you have.
The budget trap is paying $2,500 or $3,000 a month for a hosted BI-as-code workspace, or $575 a month for SSO and embeds, when a self-hosted free edition would have done the work. The quality trap is choosing a free query tool and later discovering you still need pixel-perfect reports, white-label embeds or business-user authoring. Helical Insight is ranked first because it avoids both traps, not because it is the most doctrinaire BI-as-code framework.
Related reading: Helical Insight vs Evidence · Helical Insight vs Lightdash · Helical Insight vs Apache Superset · Helical Insight vs Redash · 10 Best Open Source BI Tools
Try Helical Insight, the free starting point on this list
Open the trial, or clone Community Edition and run it on your own servers.
References
- Helical Insight Community Edition and product documentation: helicalinsight.com/community-edition/, GitHub.com/helicalinsight/helicalinsight
- Evidence docs and pricing (Core vs Studio, self-host limits, Team $2,500/month): docs.evidence.dev, evidence.dev/pricing
- Lightdash pricing (OSS vs Cloud Pro from $3,000/month, feature matrix): lightdash.com/pricing
- Preset pricing for managed Apache Superset (Starter $0 / 5 users; Professional $20/user/month annual): preset.io/pricing/
- Apache Superset 6.1.0 release (13 May 2026): github.com/apache/superset/releases
- Metabase pricing reference (updated 9 November 2026): metabase.com/pricing
- Redash GitHub repository and v26.9.0 release notes (24 September 2026); Redash authentication docs: github.com/getredash/redash, redash.io/help
- Rill pricing and “Rill Cloud vs Rill Developer” docs: rilldata.com/pricing, docs.rilldata.com
- Observable Framework GitHub (ISC license, v1.13.4 on 2 March 2026): github.com/observablehq/framework
- Grafana licensing and Grafana Cloud pricing: grafana.com/licensing/, grafana.com/pricing/
Disclosure: this roundup is published by Helical Insight, the maker of the tool ranked first. Facts about other products come from official vendor or project sources as of 20 November 2026. Prices and packaging change, so confirm with each vendor before you buy. Self-hosting always adds infrastructure and maintenance cost that vendors do not list as a license fee.
