Summary
The “Cooker Out of Memory” issue in UEFN has worsened since patch 38.00, now affecting projects that are even smaller than the previous threads on the issue, despite them being within project size and memory limits.
Projects that exceed roughly 20–40k actors are now failing to cook with either a “Cooker Out of Memory” or “Cook Timeout” error. The data size of the project has no measurable impact — the issue appears tied purely to actor count and the cook server’s inability to properly manage RAM or distribute load.
Even minimal project changes (sometimes as small as 1MB) can cause the cooker to fail, despite incremental cooking only processing new data. In some cases, rolling back to an older version of the project allows a successful cook, but stability remains inconsistent.
This problem did not exist before the Star Wars season, when large maps cooked fine, and it has grown steadily worse since. It now occurs even in stripped-down versions of projects with moderate actor counts.
Adding to the inconsistency, attempting to cook and launch a session on mobile/console can still show a “cook failed” error — yet somehow load into a fully functional cooked session. This directly contradicts expected cook behavior and suggests the backend system is failing to manage memory allocation and cook validation reliably.
Many creators now have been forced to spend hours rolling back, recooking, and rebuilding projects simply to publish content that should cook successfully on the first couple tries. This issue is slowing development, stifling creativity, and making UEFN creation extremely unstable for teams pushing the platform forward especially for me and my team who are developing Project Hyperion, which has insane growing popularity and demand from the community, despite the map having what epic would praise as flawless performance during a match.
Please select what you are reporting on:
Unreal Editor for Fortnite
What Type of Bug are you experiencing?
Project Size
Steps to Reproduce
-
Create or open a UEFN project containing a large environment with approximately 20,000–40,000+ actors.
-
Attempt to Cook or Upload the project.
Observe the following possible outcomes:
-
In most cases, the “Cooker Out of Memory” error appears during the cook.
-
If the error does not occur, the cook may instead hang for an extended period and fail with a “Cook Timeout” error.
-
Roll back to an older version of the project with fewer actors — the cook may succeed temporarily.
-
Gradually restore current project data and retry the cook. Even minor changes (as small as 1MB) can re-trigger the issue.
-
Optionally, attempt to cook on a mobile device. The cooker may still report failure, but a cooked session can sometimes still load successfully despite the reported error.
Expected Result
-
The cooker should successfully process projects that are within Epic’s performance and memory limits, even with high actor counts.
-
Cooking should complete within any time frame and not fail due to timeouts or memory allocation issues.
-
Successful cooking should be required for clients to join sessions — a failed cook should not result in a playable session despite the fact that sometimes it lets us into cooked sessions despite the error.
Observed Result
-
Projects exceeding roughly 30,000–40,000k actors frequently fail to cook, producing “Cooker Out of Memory” errors (Battle Royale maps).
-
When the out-of-memory error doesn’t occur, the cook process often hangs far longer than normal and ends in a “Cook Timeout” error.
-
This behavior significantly worsened after patch 38.00, affecting even stripped-down versions of maps with relatively low actor counts (10–20k) making it nearly impossible to load into even more simplified versions of projects now.
-
The failure appears unrelated to total data size and seems purely connected to actor count and the server’s memory handling.
-
Incremental cooking (which should only process newly added data) still fails unpredictably, even after previously successful cooks.
-
In some cases, mobile clients report cook errors yet still load into a fully functional session, contradicting expected behavior and suggesting inconsistent backend handling.
-
This issue did not occur before the Star Wars season update and has worsened substantially since patch 38.00 despite the last thread about the issue being marked as fixed for 38.00.
Platform(s)
PC, iOS, Android, Consoles
Upload an image
Additional Notes
This issue has become one of the biggest blockers for dedicated UEFN creators. It’s not a performance problem on our end — it’s a stability and reliability problem.
The creative community deeply appreciates the tools Epic has provided and the vision behind UEFN. Many of us are building large, ambitious projects meant to showcase what’s possible inside Fortnite and push the platform forward like was originally envisioned. But right now, the cook reliability issues are making that nearly impossible.
Every failed cook costs hours of troubleshooting, rollback work, and lost momentum. Many creators are reporting similar experiences, and the inconsistency, even more so now as people get more talented and ambitious — especially when a “failed cook” can still result in a functional session as I have shown many times.
We’re not submitting this out of frustration alone, we’re asking for this to be prioritized to work alongside us and our teams so the creator community can continue to push Fortnite’s creative platform forward with confidence. UEFN has incredible potential hence why we are even here making these insane experiences.
Thank you to the Epic team for continuing to listen and work with creators on issues like this. We genuinely want to see this platform succeed. And we hope you at least give some insight on the issue, especially considering the aforementioned successful sessions despite still getting errors.
Please reach out if you need any logs or project data and we will gladly provide these things to whoever inquires.
this is scheduled to be addressed in 39.00.