Two AI models, 150 analyses a month, for teams that tune SQL Server every day.
AI SQL Tuner Studio Team is a SQL Server performance tuning tool with two AI models: GPT-5.6 Luna and our premium GPT-6 Sol. It finds what’s slowing down your servers and gives you prioritized, ready-to-run fixes in about a minute. Team gives you 150 analyses a month, every reasoning level, and unlimited analyses on your own AI endpoint. All for $15 a month, with a 14-day free trial.
- GPT-6 Sol, our premium model, alongside GPT-5.6 Luna
- 150 analyses every month, 3 times Pro
- Unlimited analyses when you bring your own AI endpoint
- Low, Medium and High reasoning effort
- SQL Server, Azure SQL Database, Azure SQL Managed Instance and Fabric SQL Database
- Up to 2 server connections
What Team adds to Pro
Team includes everything in Pro, plus a second, premium model and three times the analyses.
| Pro | Team | |
|---|---|---|
| Analyses per month | 50 | 150 |
| GPT-5.6 Luna | ✓ | ✓ |
| GPT-6 Sol (premium model) | – | ✓ |
| Reasoning effort | Low, Medium, High | Low, Medium, High |
| Bring your own AI endpoint, with no monthly limit | ✓ | ✓ |
| All six tuning goals | ✓ | ✓ |
| Every SQL Server platform | ✓ | ✓ |
- A second, premium model. Pick GPT-6 Sol or GPT-5.6 Luna for each analysis. Run a hard problem through both and compare the advice before you change anything.
- Three times the analyses. 150 a month covers daily checks across your servers, regular index and code reviews, and the incidents that come up along the way.
- Everything in Pro. High reasoning effort and unlimited analyses on your own AI endpoint are included.
Meet GPT-6 Sol
GPT-6 Sol is our premium reasoning model, hosted on Microsoft Foundry in Azure. Like every model we offer, it earned its place on real benchmark workloads, where we apply each model’s advice and measure what it actually does to the workload.
- Query Tuner: with the execution plan in hand, GPT-6 Sol’s advice cut our benchmark query’s run time by 97 to 100 percent in every run.
- Index Tuning: on our OLTP benchmark with a realistic workload history, its index advice cut total workload time by a median of 44 percent.
- About a minute per report at Medium reasoning effort, about the same as GPT-5.6 Luna.
For the best results, run GPT-6 Sol at Medium or High reasoning effort. See the benchmark results.
Six tuning goals, one desktop app
Every goal collects the right diagnostics from your server, sends them to the AI, and returns an HTML report with an executive summary, prioritized findings and T-SQL scripts you can review and run.
| Goal | What you get |
|---|---|
| Query Tuner | Paste a slow query. Studio reads its execution plan and every table it touches, then returns the costliest operators, index scripts with their estimated size, and a rewritten query. |
| Index Tuning | Database-wide index advice from Query Store and usage statistics: missing, duplicate and unused indexes, each with the storage it adds or frees. Drop advice waits until the server has been up long enough to trust the numbers. |
| Server Health Check | An executive-level view of your instance: configuration, memory, waits, storage, tempdb, and patch level, with the issues that matter most first. |
| Code Review | A review of your stored procedures, functions, triggers and views, with full replacement scripts for the code that needs fixing. |
| Fix Deadlocks | Reads your deadlock history, pulls the code each session was running, and returns rewritten procedures that prevent the deadlock. |
| Locking and Blocking Analysis | Shows what is blocking what right now, why, and the changes that keep work flowing. |
See sample reports or watch the 2-minute tour.
Bring your own AI endpoint
Team can also send analyses to an AI model you run in your own Azure subscription instead of our hosted service.
- No monthly limit. Analyses on your own endpoint don’t count against your 150. Your AI provider bills you for the usage directly.
- Your data path, your rules. Studio sends the diagnostics straight from your machine to your endpoint. They never pass through AI SQL Tuner’s service, which only confirms your license is active.
- Your choice of model. Use GPT models deployed in Azure OpenAI, or Claude models deployed in Microsoft Foundry.
- Your key stays on your machine. Your API key is entered in Studio and saved locally only if you choose to. It is never sent to AI SQL Tuner.
Setup takes about five minutes: Configure your own AI endpoint in 5 steps.
Advice you can check
AI SQL Tuner Studio is built to give advice you can verify, not guesses:
- Grounded in your data. Every figure in a report is quoted from what Studio collected, or clearly labelled as an estimate with what it is based on. When data is missing, the report says so.
- Storage sizes on every index change. Know what a recommendation will cost on disk before you run it.
- Ready-to-run scripts. Reports include complete T-SQL, not vague suggestions.
- Tested on real workloads. We benchmark every model on real Index Tuning, Query Tuner and Code Review workloads and measure what the advice actually does.
How it works
- Start your free trial. Check out with your email address and a payment card. Nothing is charged today; your $15 monthly subscription starts after 14 days unless you cancel.
- Install and activate. Download Studio, click License…, and paste the license key from your purchase receipt.
- Connect and run. Connect with a least-privileged SQL login, pick a tuning goal and a model, and click Run Analysis. Your report starts streaming in within seconds.
No agents to install on your server. No log shipping. No changes to your database unless you choose to run a script.
Your data stays under your control
- You see everything before it’s sent. Before each run, Studio shows a run plan listing every collector and the system views it reads. Nothing is sent until you approve it.
- Only metadata leaves your machine. Studio sends diagnostic metadata, such as query text, execution plans, index statistics and configuration, to the AI model: our hosted models on Microsoft Foundry in Azure, or your own endpoint. Table data and query results are never sent.
- Nothing is kept by default. Each analysis is answered and discarded. You can opt in to share analyses to help improve the advice, and delete anything shared from inside the app.
- Read-only by design. Studio runs with a least-privileged login and never changes your database on its own.
Compare editions
Every edition includes all six tuning goals and every SQL Server platform. Managing more than two servers? Choose Enterprise, with unlimited connections and 400 analyses a month.
| Free | Pro | Team | Enterprise | |
|---|---|---|---|---|
| Price | $0 | $10/month | $15/month | $20/month |
| Analyses per month | 10 | 50 | 150 | 400 |
| Server connections | Unlimited | Up to 2 | Up to 2 | Unlimited |
| All six tuning goals | ✓ | ✓ | ✓ | ✓ |
| Every SQL Server platform | ✓ | ✓ | ✓ | ✓ |
| GPT-5.6 Luna | ✓ | ✓ | ✓ | ✓ |
| GPT-6 Sol (premium model) | – | – | ✓ | ✓ |
| Reasoning effort | Low, Medium | Low, Medium, High | Low, Medium, High | Low, Medium, High |
| Bring your own AI endpoint | – | ✓ | ✓ | ✓ |
| 14-day free trial | n/a | ✓ | ✓ | ✓ |
Free · Pro · Enterprise · Compare editions in detail
System requirements
- Windows 10 version 1809 or later, or Windows 11
- Internet connection for activation and analysis
- SQL Server 2016 or later, Azure SQL Database, Azure SQL Managed Instance, or Fabric SQL Database
- A SQL login with read-only diagnostic permissions (3-step setup guide)
- For bring-your-own AI: an Azure OpenAI or Microsoft Foundry resource with a deployed model
Frequently asked questions
How does the 14-day free trial work?
You get the full Team edition, including GPT-6 Sol, for 14 days. A payment card is required at checkout, but nothing is charged until the trial ends. After that, Team is billed at $15 a month, plus applicable tax, until you cancel.
What is the difference between GPT-5.6 Luna and GPT-6 Sol?
Both are reasoning models hosted on Microsoft Foundry in Azure, and both were chosen from our benchmark results. GPT-5.6 Luna is the default in every edition. GPT-6 Sol is our premium model, included with Team and Enterprise. You can switch between them for each analysis, so you can compare their advice on the problems that matter most.
Which reasoning effort should I use with GPT-6 Sol?
Medium or High. Medium is the default and works well for most analyses. Use High when a Medium report misses a root cause you can see in the data; it takes about two to three times as long.
What counts as an analysis?
Each run of a tuning goal on our hosted AI service counts as one analysis, with either model. A run that times out does not count, and runs on your own AI endpoint never count. The app shows how many analyses you’ve used and when your count resets.
What happens when I use all 150 analyses?
Hosted runs pause until your monthly count resets. You can keep tuning on your own AI endpoint, which has no monthly limit, or upgrade to Enterprise for 400 analyses a month.
Does bringing my own AI endpoint use my monthly quota?
No. Analyses on your own endpoint are unlimited and never count against your 150. Studio calls your endpoint directly, and your AI provider bills you for the usage.
What does “up to 2 server connections” mean?
Team is licensed for use with up to two SQL Server instances or Azure SQL servers. For more, choose Enterprise, which has no connection limit.
Can I change editions later?
Yes. You can move to Enterprise at any time, or switch to Pro if you no longer need GPT-6 Sol.
Is my data stored in the cloud?
No, not by default. Each analysis is answered and discarded. Studio sends diagnostic metadata to the AI for analysis, never table data or query results, and you see the full run plan before anything is sent. With your own endpoint, the diagnostics go only to your endpoint.
Start your free trial
Try AI SQL Tuner Studio Team free for 14 days: GPT-6 Sol and GPT-5.6 Luna, 150 analyses a month, and unlimited analyses on your own AI endpoint.
You must log in and be a buyer of this download to submit a review.
