Claude Code
8
Incidents
6.4
Avg Severity
3
Critical
2
High
Top Failure Modes
Destructive Action
3
Other
2
Security Vulnerability
2
Worst Incident
STUPID-2026-0003 10/10Claude Code ran rm -rf on test fixtures thinking they were temp files
Openai Operator
1
Incidents
4.4
Avg Severity
0
Critical
0
High
Top Failure Modes
Logic Error
1
Worst Incident
STUPID-2026-0059 4.4/10OpenAI's Operator scored 38% on real computer tasks — and panics instead of recovering from errors
Comparison Summary
| Metric | Claude Code | Openai Operator |
|---|---|---|
| Total Incidents | 8 | 1 |
| Avg Severity | 6.4/10 | 4.4/10 |
| Critical Incidents | 3 | 0 |
| Top Failure Mode | Destructive Action | Logic Error |
Frequently Asked Questions
Is Claude Code or Openai Operator more reliable?
Based on StupidLLM data, Claude Code has 8 documented failures (avg severity 6.4/10) while Openai Operator has 1 (avg severity 4.4/10). Openai Operator shows better reliability based on average severity scores.
What are the main differences between Claude Code and Openai Operator failures?
Claude Code's most common failure mode is destructive action, while Openai Operator most commonly fails via logic error. Claude Code has 3 critical incidents vs Openai Operator's 0.