Overview

Natural Language SQL is an open-source engine that translates natural-language database questions into SQL, executes them, and returns streamed results. It is best suited to developers and teams able to self-host and review generated queries. Its local-first AI and editable SQL are attractive; the sizable model download and default write permissions demand deliberate setup.

Key features

Hybrid schema retrieval uses BM25 alongside sentence-transformer vectors to select database context for each question. That helps connect a prompt to an existing schema without requiring users to assemble context manually. The project reports 52.32% execution accuracy on Spider dev, so generated SQL should be treated as a draft to inspect, not an answer to trust automatically.

Users can edit queries before execution, and streamed progress shows the stages of the workflow. Query explanations, sorting, 50-row pagination, CSV and JSON downloads, and tab-separated copying support inspection and follow-up analysis. SELECT queries receive a default LIMIT; explicit result counts are capped at 1,000 rows, and execution times out after 30 seconds by default. These defaults help constrain routine queries, but do not prevent writes: INSERT, UPDATE, and DELETE are permitted by default and accompanied by warnings. The project recommends using a read-only database account when writes must be blocked.

PostgreSQL and MySQL are supported, including parallel queries across multiple databases of the same type. Ollama runs locally by default, keeping questions, schema, and generated SQL on the user's machine; the project says it collects no telemetry. Opting into OpenAI, Gemini, or Groq sends the question, relevant schema portion, and SQL context to that provider, but excludes database passwords and result rows. Passwords and cloud API keys are Fernet-encrypted at rest, redacted from logs, and masked in API responses. Query results, schema metadata, and query history remain in memory for the session and are not written to disk by the application.

Pricing

The Natural Language SQL Engine plan costs 0.00 USD per free. It is a GPL-3.0 open-source project for local installation, with no paid tier or per-seat charge stated. That makes it a fit for teams prepared to operate the software themselves; the tradeoff is that deployment and local model resources are their responsibility. The recommended Docker setup requires Docker 20.10 or later with Compose v2 and about 10 GB of free disk space, and the first run downloads an approximately 5 GB model.

Platforms

Natural Language SQL supports API, Linux, macOS, self-hosted, and web use. Its local Ollama option suits teams that want database questions and generated SQL to stay on their machine, while cloud model options involve sending query context to the selected provider. GPL-3.0 licensing is relevant for teams evaluating how an open-source component fits their deployment.

Who it's for

This is a strong fit for technically capable teams with PostgreSQL or MySQL databases who want a self-hosted natural-language query workflow, local model execution, and the ability to review SQL before it runs. It is less suitable for teams seeking a managed, low-resource setup or a workflow in which write queries are blocked by default. The Docker footprint and write permissions make infrastructure and database access controls part of adoption, not afterthoughts.

Pros and cons

Pros: Local Ollama execution keeps questions, schema, and generated SQL on the user's machine.
Pros: Editable SQL, streamed execution progress, explanations, and export options support review and reuse of results.
Pros: PostgreSQL and MySQL connections can cover multiple same-type databases in parallel.
Cons: The recommended Docker deployment needs about 10 GB of free disk space and downloads an approximately 5 GB model on first run.
Cons: Write queries are allowed by default, so teams must rely on warnings and database permissions to prevent unintended changes.
Cons: Reported Spider dev execution accuracy is 52.32%, reinforcing the need to inspect generated SQL.

Alternatives

AI SQL Generators is the broader category to explore when comparing tools beyond this self-hosted engine.

NatureQuery is a reasonable alternative for teams seeking a freemium API and web product: its free plan includes 50 queries per month, one database connection, exports, and 30-day query history. Choose it over Natural Language SQL if those hosted-plan limits suit the workflow better than local installation.

Outerbase AI may fit teams that need broader platform support, including Windows, or a free plan with up to five users and one base. Its free tier is limited to 10 EZQL queries per month and transactional databases, so Natural Language SQL is a better match for teams prioritizing local model execution and self-hosting.

Wren AI offers an open-source context engine for individual developers through CLI and MCP without a UI. Choose it if that context-engine approach is preferable to Natural Language SQL's query interface and streamed results.

mnemiq is another free, self-hosted open-source engine; consider it as an alternative if you want to compare a different self-hosted option.

Text2SQL.ai is a paid alternative with a free trial and API access, a possible fit for teams that prefer a product with API usage terms rather than operating a local installation.

Vanna AI offers both API and self-hosted options, with a freemium model; consider it if you want to compare another platform spanning those deployment approaches.

Dataherald is a free API and self-hosted alternative for teams comparing open deployment options.

SQLAI.ai is a paid alternative with a free trial and web/API platforms; consider it if a paid service and trial are more appropriate than self-hosting.

Verdict

Choose Natural Language SQL if your team can self-host, wants local Ollama processing, and values the ability to inspect and edit SQL before execution. Its main advantage is combining schema-aware generation with privacy-conscious local operation at no software cost. Look elsewhere if you need a lighter deployment or write protection built in: the model consumes substantial disk space, and blocking changes requires a read-only database user.

	Natural Language SQL	Savvina AI
Free plan	Yes	Yes
Free trial	No	No
Paid from	Free	€39/mo
Open source	No	No
Platforms	api, Linux, macOS, self-hosted, Web	api, self-hosted, Web
Query explanations	Yes	—
Deployment	self_hosted	self_hosted
Schema context	Yes	Yes
Free plan	—	Yes

Natural Language SQL vs Savvina AI

Natural Language SQL

Savvina AI