How this list is organised#
Searching for "elasticsearch gui" returns the same ten names in a different order each time. The job matters more than the order, so the tools sit under six jobs. Each entry gives the licence, where it runs and the trade-off, with a link to the source; one table at the end puts them side by side.
SQL analysis and exports#
FlatQL#
FlatQL is a SQL workbench for Elasticsearch 6 to 9 and OpenSearch 1 to 2, in the browser today and as a desktop app soon (macOS first). It splits each statement: the cluster executes filters, sorts, limits and GROUP BY as composite aggregations and date histograms; an embedded DuckDB engine executes JOINs across indices, CTEs, window functions such as ROW_NUMBER and LAG, UNION, HAVING and QUALIFY on the rows returned. The result panel shows which part went where and the exact Query DSL. Exports stream to CSV, TSV, JSON, NDJSON, Excel or Parquet up to 5,000,000 rows. Teams share connections with credentials encrypted server-side and roles per index pattern enforced by FlatQL. Ask FlatQL drafts or explains SQL and DSL from the mappings; nothing runs until you press Run. Trade-off: the engine stage is limited to about 50,000 rows per index per statement, so a JOIN on two large indices needs a WHERE or a LIMIT (limits), and there is no shard or snapshot management. Free is one cluster, one worksheet, no export; Pro is a $199 one-time licence (pricing). Start with your first query.
Cluster administration#
ElasticVue#
ElasticVue is MIT licensed and runs as a desktop app (Windows, macOS, Linux), a Chrome, Firefox or Edge extension, a hosted web app, a Docker image or a self-hosted build. It manages indices, aliases, shards, snapshots and repositories, searches and edits documents, and has a REST editor. Its README promises compatibility with every Elasticsearch release, including end-of-life ones, and the project shipped v1.17.0 in October 2026. Trade-off: single-user with credentials stored locally, no documented export of search results, no mention of OpenSearch. FlatQL vs ElasticVue.
Cerebro#
Cerebro is a self-hosted admin console under the MIT licence, written in Scala and AngularJS. It needs Java 11 or newer, serves on port 9000, has official Docker images and supports LDAP authentication. Its code is organised into nodes, REST, snapshot, repository, template, index-settings, cluster-settings and _cat modules. Trade-off: the last release (v0.9.4) and last commit are from 2021. FlatQL vs Cerebro.
Elasticsearch Head#
Elasticsearch Head is Apache 2.0 and runs as a Chrome extension or a standalone Node server on port 9100, with Docker images for Elasticsearch 1, 2 and 5. The site-plugin install stopped working at version 5. Trade-off: last release April 2018, last commit November 2020, basic auth via URL parameters. The Elasticsearch Head alternatives page covers the move away from it; FlatQL vs Elasticsearch Head covers the SQL angle.
Data browsing and editing#
Dejavu#
Dejavu is an MIT-licensed web UI for Elasticsearch and OpenSearch, run from Docker on port 1358, from the hosted instance, or as a static client-side app. It browses several indices with visual filters, edits documents singly or in bulk through a Query DSL filter, imports JSON and CSV with a guided mapping, and exports a filtered view as JSON or CSV. Release 3.10.0 is from September 2025. Trade-off: no cluster administration, no aggregations or charts, and CORS must be enabled on the cluster. FlatQL vs Dejavu.
Elasticsearch Toolbox#
Elasticsearch Toolbox was an MIT-licensed Chrome app for Elasticsearch 2, archived in 2019 after Chrome apps were deprecated. It is listed because searches still surface it. Trade-off: it does not run on current Chrome and targets a cluster version that is long out of support. FlatQL vs Elasticsearch Toolbox is written for people migrating off it.
Dashboards and monitoring#
Kibana#
Kibana is Elastic's interface, self-managed or on Elastic Cloud. Its repository is triple-licensed under AGPLv3, SSPL and ELv2 with x-pack code under ELv2 only; the licensing FAQ says the default distribution is ELv2. Discover queries with KQL, Lucene or ES|QL and exports CSV within size limits; dashboards, Lens, Maps, alerting and the Observability and Security apps are why it exists, and Console is its REST editor. Role-based access control is free since 6.8 and 7.1; SSO, LDAP, field- or document-level security and PDF or PNG reports are paid. Trade-off: a cluster-side deployment, no SQL (ES|QL's LOOKUP JOIN needs a lookup-mode index), and every user needs a cluster identity. FlatQL vs Kibana.
Grafana#
Grafana is AGPL-3.0 and runs self-hosted or as Grafana Cloud. Its Elasticsearch data source targets Elasticsearch 7.17, 8.x and 9.x with Lucene queries, bucket and metric aggregations, raw Query DSL and ES|QL, and a separate OpenSearch data source covers Lucene and PPL for logs, metrics and traces. Grafana Cloud has a free tier with three active users; Pro is $19 per month plus $8 per active user. Trade-off: a dashboarding and alerting layer, not a document browser or admin console. FlatQL vs Grafana.
Database IDE users#
DBeaver#
DBeaver Community is Apache 2.0; Elastic's guide to DBeaver and Elasticsearch SQL shows the connection over the Elasticsearch SQL JDBC driver, and the paid Enterprise edition ships Elasticsearch, Open Distro and OpenSearch drivers with enhanced features on an annual subscription. You get a familiar IDE, table browsing and DBeaver's export wizards. Trade-off: everything goes through x-pack SQL, whose limitations include no array fields, sub-selects only when they flatten, and a 10,000-row cap on aggregations without a LIMIT. FlatQL vs DBeaver and FlatQL vs Elasticsearch SQL go deeper.
OpenSearch users#
OpenSearch Dashboards#
OpenSearch Dashboards is Apache 2.0, with release 3.9.0 in October 2026. It includes Discover, dashboards and Dev Tools, plus Query Workbench for SQL and PPL (read-only, downloadable results), Reporting to PDF, PNG, CSV and XLSX with a 10,000-row CSV default, and the Security plugin with index, document and field-level roles, multi-tenancy and audit logs. Trade-off: self-hosted service, and its SQL has no cross-index JOIN. FlatQL vs OpenSearch SQL and the OpenSearch GUI page explain where FlatQL adds to it.
Comparison table#
| Tool | Form | Query language | Joins | Exports | Access control | Licence / price | Best for |
|---|---|---|---|---|---|---|---|
| FlatQL | Web; desktop soon | SQL | Across indices, in engine | CSV, TSV, JSON, NDJSON, Excel, Parquet | FlatQL roles per index pattern | Free; Pro $199 once | SQL analysis, exports, teams |
| ElasticVue | Extension, desktop, web, Docker | REST, Query DSL | No | None stated | Cluster users | MIT, free | Cluster admin |
| Cerebro | Self-hosted (JVM) | REST | No | None stated | LDAP login; cluster users | MIT, free | Shard and settings admin |
| Elasticsearch Head | Extension, web | Query DSL | No | None stated | Cluster users | Apache 2.0, free | Legacy clusters |
| Dejavu | Web, Docker | Filters, Query DSL | No | JSON, CSV | Cluster users | MIT, free | Browsing and editing |
| Elasticsearch Toolbox | Chrome app (retired) | Query DSL | No | None stated | Cluster users | MIT, archived | Nothing current |
| Kibana | Self-hosted, cloud | KQL, Lucene, ES|QL | LOOKUP JOIN (lookup index) | CSV; PDF/PNG paid | Elasticsearch roles | ELv2 / AGPLv3 / SSPL; free tier | Dashboards, Observability |
| Grafana | Self-hosted, cloud | Lucene, aggregations, ES|QL, PPL | No | Panel data | Grafana users | AGPL-3.0; Cloud free tier | Dashboards across sources |
| DBeaver | Desktop | x-pack SQL (JDBC) | x-pack SQL only | DBeaver exporters | Cluster users | Apache 2.0; paid editions | SQL IDE users |
| OpenSearch Dashboards | Self-hosted | DQL, SQL, PPL | No | PDF, PNG, CSV, XLSX | Security plugin roles | Apache 2.0, free | OpenSearch clusters |
Picking one#
Three questions settle it for most teams:
- Do you need to change the cluster from the GUI? Shards, aliases and snapshots mean ElasticVue or Cerebro; editing documents means Dejavu or ElasticVue.
- Is the output a dashboard or a file? Dashboards mean Kibana, OpenSearch Dashboards or Grafana. A large file of rows means a SQL tool with a streaming export.
- Who else needs in? People who should not hold cluster credentials need a layer that owns the connection and enforces per-index roles itself. Of the ten tools here, FlatQL is built around that.
To test the SQL side, create a free account and run a grouped query with a filter against one cluster; the result panel shows what the cluster did and what the engine did. The Elasticsearch GUI page has the longer version.