# QBCore vs ESX: architecture and migration tradeoffs

QBCore and ESX are open-source FiveM roleplay frameworks with different APIs, resource ecosystems, data models, and community conventions. Neither has a universal performance advantage: actual results depend on the selected resources, custom code, database workload, artifacts, host, and player behavior.

For ESX Legacy installation and API guides, see [ESX Framework](https://esx-fivem.com/).
Verify version-specific details against the [ESX Core source](https://github.com/esx-framework/esx_core).

Comparing more than these two? Read the [full FiveM framework comparison](/tutorials/fivem-frameworks-comparison)
for Qbox, vRP, ND Core, Ox Core, Mythic and other alternatives, with compatibility and migration checks.

## Decision matrix

| Question | QBCore | ESX |
| --- | --- | --- |
| Primary source | [QBCore GitHub](https://github.com/qbcore-fivem) | [ESX GitHub](https://github.com/esx-framework) |
| Documentation | docs.qbcore.org | Project documentation linked by ESX |
| Compatibility | qb-* APIs and shared data | ESX APIs and ecosystem conventions |
| Migration impact | Requires porting ESX consumers and data | Requires porting QBCore consumers and data |

## Compare your actual stack

Inventory jobs, inventory, phone, housing, vehicles, banking, targeting, menus, voice, NUI, database schema, permissions, and paid resources. For each dependency, verify whether a maintained implementation exists for the target framework and whether its license permits migration. If you stay on ESX, the [ESX scripts on RPScripts](https://rpscripts.net/product-category/esx-scripts/) list commercial resources with declared ESX support; RPScripts and qbcore.net are operated by the same owner, so verify each listing yourself.

## Measure instead of assuming

1. Build equivalent staging profiles from recorded commits.
2. Use the same artifact, host, database, player scenarios, and measurement duration.
3. Record idle and representative `resmon`/profiler samples per resource.
4. Measure database latency, server hitch warnings, client frame time, reconnects, and failures.
5. Publish raw configuration and results before claiming a winner.

## Migration cost

Changing framework means translating events, callbacks, player/job/item shapes, permissions, persistence, and every dependent resource. A percentage “success rate” without a defined population and dataset is not useful evidence.

> **Do not combine frameworks casually**
> Running two framework cores against the same player and economy data can create conflicting ownership and persistence. Prefer explicit adapters at narrow boundaries and migrate one domain at a time.

Use the [migration planning guide](/docs/migration) and prove backup restore before committing to a framework change.
