Open Source Data Visualization Tools: the 2026 Guide, With a Reality Check

On August 17, 2026, we checked the actual state of the 27 data visualization tools presented as open source by at least two of the twelve top-ranking Google pages: license, latest release date, six-month activity. The result: 9 out of 27 — a third — don't keep the promise, including 6 that simply aren't open source and 3 that are abandoned or dead. The most recommended tool of the lot, D3.js, hasn't shipped a release since March 2024. None of those pages mention the license changes (Grafana moving to AGPL in 2021, RapidMiner going proprietary) or the shutdowns (Plotly Chart Studio, October 2025). One limit to our grid: it measures activity, not usability — it is harsh on stable, "finished" libraries that remain perfectly functional.

That finding is why we rewrote this guide from scratch. At 10h11 we deploy and maintain these tools in production for our clients: this is the comparison we wish we had found — verified, dated, and organized by use case instead of a flat list.

First, what is not open source (despite what you've read)

Roundups copy each other, errors included. Setting the record straight, license by license:

• Google Charts: proprietary (called "open source" by half the top 10). Free ≠ free software: no self-hosting, no forking.
• Tableau Public and Looker Studio: free, proprietary.
• Datawrapper: proprietary SaaS; the open source core has been abandoned since 2018. An excellent editorial tool — but not open source.
• RapidMiner: last free version June 2020; proprietary since the Altair acquisition.
• wpDataTables Lite: GPL on the surface, closed development.
• And two dead projects still recommended in 2026: Charted (archived, last release 2015) and Candela (last release 2018).

If a roundup lists any of these as "open source", you know what the rest of the list is worth.

The durability grid: 27 tools, measured August 17, 2026

For each tool: actual SPDX license, latest release, six-month activity, governance. Verdicts: active (release < 6 months), slowing (6-18 months), zombie (> 18 months), dead (archived), not OSS.

Tool	License	Latest release	Verdict
Apache ECharts	Apache-2.0	May 2026	Active
Apache Superset	Apache-2.0	May 2026	Active
Bokeh	BSD-3	Jul. 2026	Active
Chartist.js	MIT	Sep. 2025	Slowing
Cytoscape.js	MIT	Aug. 2026	Active
D3.js	ISC	Mar. 2024	Zombie*
dygraphs	MIT	Jul. 2026	Active
Gephi	GPL-3.0	May 2026	Active
ggplot2	MIT	Apr. 2026	Active
Grafana	AGPL-3.0 (ex-Apache, 2021)	Aug. 2026	Active
KNIME	GPL-3.0 + exception	Jul. 2026	Active
Leaflet	BSD-2	May 2023 (v2 alpha)	Slowing*
Matplotlib	Matplotlib license	Jul. 2026	Active
Metabase	AGPL-3.0 + commercial ed.	Jul. 2026	Active
Plotly / Dash	MIT	Jul. 2026	Active
RAWGraphs	Apache-2.0	Jan. 2024	Slowing
Redash	BSD-2	Mar. 2026	Active (community-run)
Seaborn	BSD-3	Jan. 2024	Slowing*

* Important nuance: "zombie" and "slowing" measure activity, not quality. D3.js, Leaflet and Seaborn are stable, finished libraries — perfectly usable, but you should know you're adopting a tool that no longer evolves, not an actively developed project. Leaflet, for instance, remains very active commit-wise (22 contributors over six months) despite a 2023 stable release.

Four major events that no top-10 roundup mentions: Grafana's move to AGPL (2021), RapidMiner going proprietary, Redash freezing for three years after Databricks acquired its team — the project was revived by volunteers and is alive again since 2026 — and the shutdown of Plotly Chart Studio (October 31, 2025).

Governance predicts the paywall

The clearest pattern in our data: with single-vendor tools (Metabase, Grafana), the same trio of features is consistently gated behind paid editions — SSO/SAML, row-level security (RLS), scheduled reports. With foundation-governed tools (Superset and ECharts under Apache; Matplotlib and Bokeh under NumFOCUS), nothing is paywalled. This isn't a moral judgment: it's a practical selection criterion. If your requirements include SSO or scheduled report delivery, a single vendor's "free" probably won't stay free for you.

Which tool for which job: our matrix

Self-hosted internal BI (the most common case among our clients). Apache Superset if you have a team to operate it: the most complete option, foundation governance, nothing paywalled. Metabase if simplicity wins: far faster to get running, but AGPL-licensed with RLS/SSO in the paid tier. Redash is alive again, but adopt it knowing its history.

Monitoring and technical real-time. Grafana, no debate — that's its home turf. For business BI, though, here's the straight answer to a question we hear constantly: no, Grafana is not a BI tool. It excels at time series and infrastructure metrics; the moment you need pivot tables, business drill-downs or sharing with non-technical users, Superset or Metabase are the right tools. Roundups that file Grafana under "BI tools" keep the confusion alive.

Visualization embedded in a product. The real gap in the open source market — we get asked about it constantly, and no roundup covers it. Two strategies: libraries (Apache ECharts, badly underrated in roundups — two mentions across twelve pages, yet it's our default choice for embedded work: Apache-2.0, fast canvas rendering, themeable; Plotly.js; D3 when you need absolute custom control) or the BI platforms' embedded modes — knowing that serious multi-tenancy (per-customer RLS) is precisely the paywalled feature at single-vendor shops.

Editorial and one-off visualization. RAWGraphs to explore without code (development has slowed but it works). And honestly: if open source is not a hard constraint, Datawrapper remains an editorial-quality reference — just know it isn't open source.

Data applications in Python/R. Streamlit, Dash, Panel, Voilà on the Python side, Shiny for R: the application layer above visualization, turning an analysis into an interactive tool. It's often the real answer when a client says "we need a dashboard" and the need is actually a business tool.

The questions people actually ask (that roundups dodge)

"What does self-hosting really cost?" The most asked question on Reddit and Hacker News, never answered with numbers anywhere. The honest answer: the free license shifts the cost to engineering — server sizing, updates, metadata database backups, version upgrades. The order of magnitude depends mostly on two choices: tool complexity (Superset demands significantly more operational attention than Metabase) and your availability requirements.

"Do updates break dashboards?" The classic trap: the metadata database. An instance set up "just to try it" on the default embedded database loses logins and dashboards at the first serious migration. Golden rule from our deployments: external metadata database (PostgreSQL) from day one, backed up before every upgrade, and read the release notes — Superset's major migrations are not skippable.

"What about scheduled report delivery?" A truth rarely written down: at most single-vendor tools, scheduled delivery (email, let alone SharePoint/FTP) is reserved for paid editions. If this is a core need, check that exact point before choosing — or plan to build it alongside.

Methodology

The list of 27 tools was frozen before measurement: every tool cited by at least two of the twelve top-10 Google pages for "open source data visualization tools" (August 2026 sample). Measurements taken August 17, 2026 via the GitHub API (SPDX license, latest release, activity, archive status), completed with official announcements for license changes. A tool without a verifiable public repository is classified by its actual license, not its reputation. Our possible bias: we're an agency that deploys these tools — our interest is in self-hosting done well, not in self-hosting being chosen.

Torn between two tools for your context? That's literally our job — let's talk.