SQLly vs DBDelve
DBDelve is a GPL-3.0 database client written in Rust on
GPUI - the same UI framework SQLly uses, down to the same gpui-pre version. It
started in August 2026, ships a release every few days, and already covers 11 engines plus an
ODBC bridge, with nothing behind a paywall. SQLly is built on the same foundations and spends
its extra weight on IntelliSense, guardrails, and DBA tools. Here’s the honest breakdown.
Last reviewed: November 2026
The short version
A deep SQL workbench
One Rust codebase drawn on the GPU on macOS, Windows, and Linux. Thirty engines, IntelliSense that reads your data, layered safety guardrails, plans, compare, AI, an MCP server, and a full set of script-first admin tools. Signed and notarized on the Mac.
A fast, free client that stays out of the way
Also Rust and GPUI. Eleven built-in engines including MongoDB and Snowflake, an ODBC bridge for the rest, per-connection read-only and full-access modes, a plan viewer, themes you can import from Zed or VS Code, and a launch that reopens your last results. Young: version 0.7.0, about 32 releases since August, 13 contributors.
Feature by feature
SQLly’s column uses the same honest status badges as the features page and the docs.
For DBDelve we list what its README, website, docs folder and Cargo.toml state, as of v0.7.0 (November 5, 2026).
| Capability | SQLly | DBDelve |
|---|---|---|
| How the app is built | Rust + GPUIOne Rust codebase, rendered on the GPU (Metal, Direct3D, Vulkan). How it works | Rust + GPUIThe same framework and version (gpui-pre 0.3.9), with gpui-component vendored and patched for a frosted-glass blur. |
| Runs on macOS, Windows & Linux | YesmacOS is signed and notarized; Windows and Linux are a supported beta, on amd64 and arm64. | YesmacOS 12+, Windows x86_64, Linux x86_64 and aarch64 (tarball, AppImage, AUR). |
| Signed Mac build | YesDeveloper ID signed and notarized, plus a Mac App Store edition. | NoAd-hoc signed; you clear the quarantine flag after each install, and macOS asks again for Keychain passwords after each update. |
| Intel Macs | NoApple Silicon only. | NoApple Silicon builds only; Intel is described as untested rather than blocked. |
| Download size | LargermacOS dmg 65.6 MB, Windows zip 60.9 MB, Linux tarball 76.2 MB. | Smallerv0.7.0: macOS dmg 54.2 MB, Windows zip 45.3 MB, Linux x86_64 tarball 61.7 MB. |
| Free to use | Free tierTwo saved servers, one cloud account, and a reminder to pay about every five hours. Pricing | YesEverything, with “no feature ever going behind a paywall”. No account, no telemetry. |
| Open source | NoLifetime licenses carry a source-escrow promise if SQLly is ever discontinued. | YesGPL-3.0 on GitHub (earlier contributions MIT). |
| PostgreSQL, MySQL, MariaDB, SQLite, SQL Server | YesPlus Azure SQL (Entra sign-in in progress) and hosted compatibles - 30 engines in all. Full list | Yes |
| DuckDB, ClickHouse, Redis | YesRedis with Cluster, Sentinel, Valkey, KeyDB, DragonflyDB, ElastiCache and MemoryDB. Redis | YesDuckDB embedded; ClickHouse over HTTP only (no native port 9000); Redis with Valkey, KeyDB and Dragonfly. |
| libSQL / Turso | YesSelf-hosted sqld or Turso Cloud over HTTPS. |
PartialLocal Turso files. |
| Oracle Database | YesNative driver, with plans and admin tools. | PartialThrough the ODBC bridge with Oracle’s ODBC driver; grid editing is off over ODBC. |
| MongoDB | No | Yesmongosh-style statements on the official Rust driver. |
| Snowflake | Planned | YesOver Snowflake’s SQL REST API, with key-pair auth. |
| ODBC for anything else | NoEvery engine is a native driver. | YesA bundled ODBC bridge (Db2, Informix, Teradata, Firebird…); bring the vendor’s driver. Windows ODBC is marked untested. |
| SSH tunnels | YesSSH hops, SOCKS and HTTP proxies, and Kubernetes. Engine docs | YesRuns your own ssh, so your config, agent and hardware keys apply. |
| Import connections from other tools | YesDataGrip, DBeaver, pgAdmin, Azure Data Studio, DbGate, TablePlus, Navicat, Redis Insight; exports too. Passwords aren’t carried. Compatibility | YesDBeaver (all platforms) and TablePlus (macOS), passwords and SSH settings included. |
| Autocomplete | YesAlias- and scope-aware, JOIN ON from foreign keys, real column values in WHERE. IntelliSense |
PartialTables, and a table’s columns after its name and a dot, from the loaded catalog. |
| Command palette | YesCovers the whole menu bar. | YesOne fuzzy palette for commands, tables, saved queries, history and connections, and every shortcut rebindable. |
| Themes | YesClassic, Neutral and Signature families, light and dark. | YesMore of them: Catppuccin, Tokyo Night, Gruvbox, Dracula, translucent glass variants, and import from Zed or VS Code. |
| Edit table data | YesA row editor generated from the table’s own metadata. Row editor | YesEdit a cell and the UPDATE lands in the editor for you to read and run; a row panel for long and JSON values. |
| Large results | YesStreams and spills to disk past 50,000 rows or 64 MiB, so memory stays flat at 50 million rows. Streaming | YesA row limit by default; with it off, its own benchmark held 1 million rows (4 GB of data) in 4.9 GB of memory. |
| Pivot, charts & result diff | YesPivot, charts, conditional formatting, computed columns, and diffs on any result. Pivot | No |
| Import & export data | YesCSV, JSON, Excel with live formulas, SQL, and more; cross-engine Data Transfer. | PartialExport CSV, TSV, JSON or INSERTs; import rows from CSV, TSV or JSON; drop a CSV, Excel, Parquet or JSON file to query it in DuckDB. |
| Execution plan viewer | In progressPlain-language findings, tree, and diagram across nine engines. Query plans | YesA highlighted plan pane that calls out bad row estimates and wasteful filters. |
| Guardrails | YesSELECT-only connections, no UPDATE/DELETE without WHERE, auto-rollback, tenant checks, inherited by environment. Safety |
PartialRead-only, read-write or full-access per connection, a colored title bar, and a prompt before anything destructive. |
| ER diagrams | YesExport to SVG, Mermaid, or DBML. Relationship Graph | NoA structure tab lists columns, indexes, constraints and foreign keys both ways. |
| Schema & data compare | YesRisk-graded schema diffs and table diffs, with sync scripts. Schema Compare | No |
| Activity monitor & admin tools | YesActivity Monitor, Server Dashboard, Query Insights, SQL Agent, backup/restore and more. Tools | No |
| AI assistant & MCP server | YesLocal models or your own endpoint, cloud opt-in; an MCP server whose reads are always rolled back. Read-only enforcement | NoNone in the app or the repository. |
| Plugins | No | PartialOut-of-process engine plugins over a shared protocol; ODBC is the first one. |
DBDelve ships fast, so a “No” here may already be out of date - check its repository for the current state.
Where SQLly is different
Same framework, same driver crates for most engines. The difference is how far each app goes past the grid.
Autocomplete that knows your data
DBDelve completes table names and the columns after table.. SQLly tracks aliases and scope, writes a join’s ON clause from real foreign keys, pulls a column’s distinct values into WHERE, and underlines a GROUP BY mistake as you type.
Guardrails beyond a mode switch
Both apps have read-only connections and ask before destructive statements. SQLly adds a hard stop on UPDATE/DELETE with no WHERE, blocks file-reading and shell-escape tricks on SELECT-only connections, wraps changes in auto-rollback, and lets a whole environment inherit those rules.
Results that don’t need a row limit
SQLly spills large results to disk instead of holding them in memory, so a 50-million-row result stays flat on RAM. On top of the grid: pivot, charts, conditional formatting, computed columns, result diffs, and Excel export with a pivot sheet.
Spill-to-disk results →DBA tools
Activity Monitor with blocking chains, Query Store and pg_stat_statements insights, Extended Events, SQL Agent, schema and data compare, an index analyzer, and backup/restore as reviewable scripts. DBDelve doesn’t try to cover this ground yet.
Tools →AI that can stay on your machine
An assistant on local models (Ollama, MLX), your own endpoint, or a cloud provider you opt into. Agents reach SQLly through sqlly-cli mcp, which runs reads in a transaction that is always rolled back and turns writes into proposals you confirm.
Native Oracle, Azure SQL, remote Turso
SQLly’s Oracle, Azure SQL (Entra sign-in in progress) and libSQL/Turso Cloud connections are native, with the editor and tools working on them. On DBDelve, Oracle goes through ODBC with editing off, and Turso opens local files.
Engines →When DBDelve is the better choice
DBDelve is probably the better fit if:
- You want everything free, at any company size. DBDelve is GPL-3.0 with no paid tier and no telemetry. SQLly’s source is closed, and its free tier isn’t available inside organizations with more than 100 employees.
- You need MongoDB or Snowflake. DBDelve has both built in; SQLly has neither (Snowflake is planned).
- Your database only has an ODBC driver. Db2, Informix, Teradata or Firebird are reachable from DBDelve’s ODBC bridge. SQLly has no ODBC.
- You want a smaller, simpler client. DBDelve’s downloads are 11 to 16 MB smaller than SQLly’s, and it keeps to browsing, querying and editing. If you won’t use plans, compare or admin tools, you don’t need to carry them.
- You care about themes. Glass themes and importing your Zed or VS Code theme are things SQLly doesn’t do.
- You want to change the code. It’s an open Rust codebase taking outside pull requests, with an engine plugin protocol.
Caveats in the other direction: DBDelve is less than three months old and pre-1.0, its Mac build isn’t notarized, and its own docs mark Windows as the least tested platform.
Questions
Is DBDelve free?
Yes. DBDelve is free and open source under the GPL-3.0, needs no account, sends no telemetry, and its author says no feature will ever go behind a paywall. SQLly is free for hobbyists, with SQLly 1.x and Lifetime licenses for everyone else.
Which databases do SQLly and DBDelve support?
Both connect to PostgreSQL, MySQL, MariaDB, SQLite, DuckDB, SQL Server, ClickHouse and Redis, including Valkey, KeyDB and Dragonfly. DBDelve adds MongoDB and Snowflake, opens local Turso files, and reaches other databases such as IBM Db2 or Oracle through a bundled ODBC bridge once you install the vendor’s ODBC driver. SQLly adds a native Oracle driver, Azure SQL (Entra sign-in in progress), remote libSQL/Turso, and documented wire-compatible services such as CockroachDB, Neon, Supabase and Amazon ElastiCache - 30 engines in all.
Are SQLly and DBDelve built the same way?
Closer than any other pair on this site. Both are Rust apps drawn on the GPU by Zed’s GPUI framework, both pin gpui-pre 0.3.9 with gpui-component widgets, and both use the tokio-postgres, tiberius, rusqlite, duckdb and redis crates for those engines. The difference is what sits on top: DBDelve is a fast, keyboard-first client; SQLly adds IntelliSense from your data, editor guardrails, AI, an MCP server and DBA tools.
Other comparisons
Point both at the same database
Both are Rust, both start fast, and both are free to try with no account. Run your own queries in each and keep the one that fits. SQLly is free on macOS, Windows, and Linux.