Dory Has Upgraded Its Managed AI Model to GPT-5.4 Mini
Dory has upgraded the model behind its managed AI service from GPT-4o mini to GPT-5.4 mini.
This update is part of our ongoing work to make Dory's AI features faster, more useful, and more reliable in everyday database workflows. When you ask Dory to generate SQL, explain a query, fix an error, summarize schema context, or reason through a result set, the managed AI path now uses the newer GPT-5.4 mini model.
Why We Made the Switch
AI is most useful in a SQL workspace when it can stay close to the work you are already doing.
Database tasks are rarely isolated prompts. A good assistant needs to understand the selected connection, visible schema, current SQL, execution errors, result shape, and the database dialect you are using. It also needs to respond quickly enough that using AI feels like part of the workflow instead of a separate research step.
Moving from GPT-4o mini to GPT-5.4 mini gives Dory a stronger foundation for those moments:
- More helpful SQL generation from natural language questions.
- Better query explanation when reading unfamiliar SQL.
- More useful suggestions when fixing failed queries.
- Smoother follow-up conversations in AI Chat.
- A better default experience for users who rely on Dory's managed AI service.
What Users Should Notice
For most users, there is nothing new to configure. The upgrade applies to Dory's managed AI service automatically.
You should simply see a better experience when using AI across Dory:
- Ask AI can produce more useful first drafts.
- Copilot actions can give clearer rewrites, fixes, and explanations.
- SQL Console assistance can better follow the query and schema context in front of you.
- Result and table summaries can feel more natural when moving from exploration to analysis.
The goal is not to make AI a separate feature you have to manage. The goal is to make it a more capable layer inside the database workspace you already use.
Built for Real Database Workflows
Dory's AI features are designed around database context, not generic chat alone.
That means the model upgrade matters most in practical workflows:
- Turning a business question into a SQL draft.
- Adapting a query to PostgreSQL, MySQL, SQL Server, Oracle, ClickHouse, SQLite, DuckDB, and other supported databases.
- Explaining complex joins, filters, aggregations, and window functions.
- Fixing SQL after the database returns an error.
- Summarizing tables, columns, and query results so you can decide what to do next.
GPT-5.4 mini gives these flows a stronger model while keeping the experience lightweight enough for frequent use.
No Action Required
If you use Dory's managed AI service, the upgrade is already part of the service path.
Self-hosted teams and organizations using their own AI providers can continue to manage their provider and model choices through their existing configuration. This announcement is about the managed model used by Dory's hosted AI service.
What Comes Next
We will keep improving the AI experience in Dory around the places where it helps most: schema understanding, SQL generation, query repair, result interpretation, and context-aware database work.
The move to GPT-5.4 mini is one step in that direction. Dory should feel more capable when you need help, and stay out of the way when you already know exactly what to run.