SQLly vs Tablething
Tablething is a local-first desktop client built with Tauri, shown on Hacker News in June 2026. It goes wide: 31 sources from Postgres to MongoDB, Snowflake and Excel files, AI on your own key, a read-only MCP server, and dashboards of saved queries, for a one-time $49 after a 7-day trial. SQLly is a closed-source workbench that goes deeper on SQL, 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, with no browser engine. Thirty engines, IntelliSense that reads your data, layered safety guardrails, schema compare, and a full set of script-first admin tools.
One paid app for many kinds of data
A Tauri 2 app: a Rust back end with its interface in the system webview. SQL, document, search, key-value and warehouse sources plus local files in one window, plain-English queries and agentic chat on your own AI key, live dashboard tiles, and an MCP server. Closed source, version 0.16.4 (September 2026), from one developer.
Feature by feature
SQLly’s column uses the same honest status badges as the features page and the docs. For Tablething we list what its website, README and release assets state. Its source isn’t public, so we couldn’t check behavior in code.
| Capability | SQLly | Tablething |
|---|---|---|
| Runs on macOS, Windows & Linux | YesmacOS is signed and notarized; Windows and Linux are a supported beta, on amd64 and arm64. | YesA signed, notarized universal .dmg, .exe and .msi for Windows x64, and .deb, .rpm and AppImage for Linux x64. The app updates itself. |
| Intel Macs | NoApple Silicon only. | YesOne universal binary for Intel and Apple Silicon, macOS 10.15 or later. |
| How the app is built | NativeOne Rust codebase, rendered on the GPU (Metal, Direct3D, Vulkan). How it works | TauriTauri 2 with a Rust back end (sqlx, tiberius, DuckDB, MongoDB and Scylla drivers, by the library names in its binary); the interface runs in the system webview. |
| Download size | MixedmacOS dmg 65.6 MB, Windows x64 zip 60.9 MB, Linux x64 tarball 76.2 MB. | Mixedv0.16.4: universal dmg 76.4 MB, Windows setup .exe 22.9 MB (.msi 33.1 MB), .deb 41.9 MB (AppImage 116.8 MB). |
| Free to use | Free tierTwo saved servers, one cloud account, and a reminder to pay about every five hours. Pricing | 7-day trialEverything unlocked for 7 days, no account. Its README badge and one comparison page still say free or beta; the homepage pricing says trial, then a license. |
| Paid license | One-time$100 for SQLly 1.x, or $200 Lifetime. Pricing | One-time$49 for one device or $99 for three, with lifetime updates and an offline license key. |
| Open source | NoLifetime licenses carry a source-escrow promise if SQLly is ever discontinued. | NoThe GitHub repository holds a README, issue templates and release files. On Hacker News its maker wrote that the plan is to open-source the core once it’s stable. |
| PostgreSQL, MySQL, MariaDB, SQL Server, SQLite | YesPlus Azure SQL (Entra sign-in in progress). Full list | Yes |
| DuckDB, ClickHouse, Oracle, Turso | YesNative drivers; DuckDB is bundled. Per-engine coverage | YesPlus MotherDuck and Cloudflare D1. Oracle needs Oracle’s Instant Client installed separately. |
| CockroachDB, Supabase, Neon & other wire-compatibles | YesDocumented wire-compatibles over the PostgreSQL and MySQL drivers, with per-engine caveats. Compatibles | YesAnything that speaks the Postgres, MySQL or Elasticsearch wire, such as TimescaleDB, PlanetScale and SingleStore. |
| Redis | YesTyped key editing, Cluster and Sentinel, pub/sub, profiler, memory analyzer; Valkey, KeyDB, DragonflyDB, ElastiCache and MemoryDB too. Redis | YesListed as a source; we couldn’t confirm which Redis tools it has. |
| Snowflake, BigQuery, Redshift & Databricks | PlannedSnowflake and BigQuery are modeled in SQLly’s engine profiles; live connectors still to come. | Yes |
| MongoDB, Elasticsearch, DynamoDB & Cassandra | No | YesPlus OpenSearch. |
| CSV, Parquet, JSON & Excel files | PartialRead Parquet, CSV and JSON offline through DuckDB’s read_parquet, read_csv and read_json. No Excel file as a source. DuckDB |
YesEach file type is its own connection you query with SQL; file connections are read-only. |
| SSH tunnels | YesSSH hops, SOCKS and HTTP proxies, and TLS modes. TLS | YesIncluding databases behind a bastion host. |
| Saved passwords | OS keychainConnection files are plain text; passwords live in the keychain. | -Stored locally; we couldn’t confirm whether it uses the OS keychain. |
| Import connections from other tools | YesDataGrip, DBeaver, pgAdmin, TablePlus, Navicat, and more. Compatibility | -Not mentioned on its site or README. |
| Autocomplete | YesAlias- and scope-aware, JOIN ON from foreign keys, real column values in WHERE. IntelliSense |
YesAutocomplete and syntax highlighting are listed; we couldn’t check how far completion goes. |
| Saved queries & history | YesA searchable execution log with pins, plus an SQL library that can be a git repository. Query history | YesSaved queries and a searchable query history. |
| Edit table data | YesA row editor generated from the table’s own metadata. Row editor | YesFilter, sort and edit in a spreadsheet-style grid. |
| Charts & pivot | YesPivot, charts, conditional formatting, computed columns, and diffs. Pivot | ChartsCharts from query results. No pivot or result diff that we could find. |
| Dashboards | NoCharts sit on a result; there’s no dashboard canvas. The Server Dashboard shows server health, not your data. | YesPin saved queries to dashboards as auto-refreshing tiles. |
| Import & export data | YesCSV, JSON, SQL, and Excel with live formulas and a pivot sheet. Export | -Copy options exist for chat results; we couldn’t confirm file import or export formats. |
| Large results | YesStreams and spills to disk past 50,000 rows or 64 MiB; flat memory at 50 million rows. Streaming | -We couldn’t confirm how large results are held. |
| Execution plan viewer | In progressPlain-language findings, tree, and diagram across nine engines. Query plans | -AI can explain or optimize a query; we found no plan viewer. |
| Schema & data compare, data transfer | YesRisk-graded schema diffs with sync and rollback scripts, data compare, and cross-engine transfer. Schema Compare | -Not on its site or README. |
| Monitoring & admin tools | YesActivity Monitor with blocking chains, Server Dashboard, backup and restore, SQL Agent, index analyzer. Tools | -Not on its site or README. |
| ER diagrams | YesExport to SVG, Mermaid, or DBML. Relationship Graph | -The sidebar shows relationships; we found no diagram. |
| Guardrails in the editor | YesSELECT-only connections, no UPDATE/DELETE without WHERE, auto-rollback, tenant checks, prod in red. Safety |
-Its MCP access is read-only; we couldn’t confirm read-only connections or other checks in the editor. |
| AI assistant | YesLocal models (Ollama, MLX), your own endpoint, or opt-in cloud providers; no default provider. On-device AI | YesYour own key for OpenAI, Anthropic, Google Gemini or Groq, or Ollama and OpenAI-compatible endpoints. Writes, explains, optimizes and fixes SQL, plus an agentic chat. |
| MCP server | YesAgent queries pass the same select-only checks as the app. Select-only | YesBuilt in, with read-only access for Claude, Cursor, Codex, VS Code, Zed and other clients, without handing over credentials. |
A dash means we couldn’t confirm it from Tablething’s own materials - check its site for the current state.
Where SQLly is different
Tablething puts many kinds of data in one window and lets AI do a lot of the typing. SQLly spends its effort on the SQL and the work around it.
No webview between you and the grid
SQLly draws its own UI on the GPU, and large results stream to disk, so the grid stays flat on memory whether you fetch 50 rows or 50 million. A 0.5 MB script still edits at about 5 ms per keystroke.
Spill-to-disk results →Autocomplete that knows your data
Beyond table and column names: SQLly pulls a column’s real distinct values into a WHERE clause, writes a join’s ON clause from actual foreign keys, reads plain-language dates, and flags a GROUP BY mistake as you type.
Guardrails beyond the AI
Read-only connections that refuse file-reading and shell-escape tricks, a hard stop on UPDATE/DELETE with no WHERE, auto-wrapped transactions, and prod painted red - inherited down your client, project, and environment tree.
AI that can stay on your machine
SQLly has no default AI provider. Run a local model through Ollama or MLX, point it at your own endpoint, or opt in to a cloud provider, and agents only ever read: writes come back to you as proposals.
On-device AI →Compare, transfer, and admin tools
Schema compare across engines with sync and rollback scripts, data compare, cross-engine data transfer with type mapping, SQL Agent, Extended Events, an index analyzer, and a test data generator.
Tools →A full Redis toolset
Typed key editing, Cluster and Sentinel, pub/sub, a profiler and a memory analyzer, plus Valkey, KeyDB, DragonflyDB, ElastiCache and MemoryDB, next to the SQL engines both apps share.
Redis →When Tablething is the better choice
Tablething is probably the better fit if:
- Your data isn’t all SQL. MongoDB, Elasticsearch, OpenSearch, DynamoDB and Cassandra sit next to Postgres in one app. SQLly connects to none of them.
- You need Snowflake, BigQuery, Redshift or Databricks today. Tablething lists all four; SQLly’s Snowflake and BigQuery connectors are still planned.
- You want dashboards. Pin saved queries as tiles that refresh themselves. SQLly charts a single result and has no dashboard canvas.
- You query spreadsheets and data files. Excel, CSV, Parquet and JSON files open as their own sources. SQLly reads the last three through DuckDB functions and doesn’t open Excel files.
- You’re on an Intel Mac. Tablething ships a universal Mac build; SQLly needs Apple Silicon.
- You want to pay less. $49 one-time for one device, or $99 for three, with lifetime updates, against SQLly’s $100 for 1.x or $200 Lifetime.
- You’d rather ask than write SQL. Its chat writes and runs queries from plain questions on the AI key of your choice.
One honest caveat in the other direction: Tablething is a young, closed-source app from one developer. Its latest release, v0.16.4, came out on 12 September 2026, and GitHub keeps only that one release, so check its site for recent activity before you rely on it.
Questions
Is Tablething free?
Not after the trial. Every download starts a 7-day free trial with everything unlocked and no account. After that Tablething costs $49 one-time for one device or $99 for three, with lifetime updates. Its README and one of its comparison pages still call it free or in beta, so check its pricing section before you plan around it. SQLly is free for hobbyists (two saved servers), with a one-time $100 SQLly 1.x license and a $200 Lifetime license for everyone else.
Which databases do SQLly and Tablething support?
Both connect to PostgreSQL, MySQL, MariaDB, SQL Server, SQLite, DuckDB, ClickHouse, Oracle, Turso, Redis and CockroachDB, plus hosted Postgres such as Supabase and Neon. Tablething adds MongoDB, Elasticsearch, OpenSearch, DynamoDB, Cassandra, Snowflake, BigQuery, Redshift, Databricks, MotherDuck and Cloudflare D1, and opens CSV, Parquet, JSON and Excel files as sources - 31 in its list. SQLly connects to 30 engines, relational, analytical and Redis, and goes deeper on each.
Do SQLly and Tablething have AI and an MCP server?
Both do. Tablething uses your own key for OpenAI, Anthropic, Google Gemini or Groq, or a local model through Ollama or any OpenAI-compatible endpoint, and its built-in MCP server gives tools such as Claude and Cursor read-only access to your databases. SQLly works with local models (Ollama, MLX), your own endpoint, or opt-in cloud providers, and its sqlly-cli mcp lets an agent read schemas and run a single SELECT in a read-only transaction that is always rolled back; writes come back to you as proposals.
Other comparisons
Point both at the same database
The quickest way to choose is to run your own queries in each. SQLly is free on macOS, Windows, and Linux, with no account required.