*NOTE This started MONDAY May 5 - i just confirmed that this issue has been resolved as of May 9…
Summary: ChatGPT Memory Crash Issue (May 4–May 9, 2025)
| Item | Details |
|---|---|
| Issue | Memory crashes when attempting to open Excel (.xlsx) or large CSV files inside ChatGPT |
| Start Date | Late night, May 4, 2025 |
| Symptoms | - Uploads worked, but |
- Attempting to open any uploaded file immediately caused a system reset/crash|
|Duration|❗ 5 days (May 4 → May 9)|
|Impact|- All Excel/CSV processing broken - File-based workflows interrupted
- Partial session memory resets wiped file context between uploads|
|Public Status from OpenAI|❌ None — no public announcement or status post|
|User Escalation Options|❌ No direct escalation path (feedback options removed from ChatGPT app and web Help Center for standard users)|
|Resolution|✅ Resolved silently by OpenAI sometime on May 9 morning (ET) — memory loading is now working again|
|Confirmed Fix|Opening the same Excel file that had caused crashes now works normally without system resets|
Updated Situation (May 7, 2025):
| Item | Status |
|---|---|
| Current Date | May 7, 2025 |
| Server memory load problem |  Still ongoing (now officially into Day 3) |
| Ability to open Excel/CSV |  Broken — still crashing memory |
| Uploads |  Still working |
| Text conversations | Still working |
| OpenAI Public Acknowledgement | None (still no posted update) |
| User Escalation Path | Still blocked for regular users |
My Original Post:
heads up - ChatGPT is warning me today about a “temporary server memory crash issue”
(Text from ChatGPT below)
What I can tell you:
- Starting earlier today (around 6–7 AM Eastern Time May 5),
many memory-intensive tasks (like opening Excel files, large CSVs, or generating big tables) started failing. - Instead of the usual smooth processing, tasks are hitting internal timeout errors or out-of-memory errors.
- This isn’t limited to you — other users are seeing similar behavior (but there’s no public dashboard for this).
- It’s almost certainly tied to temporary load balancing or server updates happening internally.
Why there’s no public announcement:
- OpenAI (the provider behind this system) usually only posts public notices when there is:
- A full system outage (site not loading at all)
- Major model unavailability (e.g., GPT-4 or GPT-3.5 entirely down)
- Partial degradations (like memory handling issues) are often just monitored internally and fixed silently.
Good news:
Typically, issues like this self-resolve within 12–24 hours without any user action needed.
Best guess:
It should stabilize overnight and be back to normal tomorrow morning.