Instrument a Java client-server application with minimal OTEL tracing. Client makes two workflows (valid token, invalid token). Must produce exactly 2 trace IDs.
Common failure modes
- Test expects exactly 2 trace IDs, models produce 3 - Java OTEL SDK complexity - Models create extra traces for initialization/startup
Performance
| Model | Pass Rate | Runs | Avg Cost | Avg Time |
|---|---|---|---|---|
| gpt-5.2-codex | 0% | $0.00 | 121m | |
| grok-4.1-fast | 0% | $0.10 | 16m | |
| gemini-3-flash-preview | 0% | $0.16 | 5m | |
| kimi-k2-thinking | 0% | $0.16 | 21m | |
| deepseek-v3.2 | 0% | $0.21 | 23m | |
| glm-4.7 | 0% | $0.23 | 12m | |
| gpt-5.1 | 0% | $0.47 | 18m | |
| claude-haiku-4.5 | 0% | $0.53 | 10m | |
| gemini-3-pro-preview | 0% | $0.57 | 10m | |
| claude-sonnet-4.5 | 0% | $0.64 | 7m | |
| gpt-5.2 | 0% | $0.72 | 17m | |
| gpt-5.1-codex-max | 0% | $1.00 | 25m | |
| claude-opus-4.5 | 0% | $1.15 | 9m | |
| grok-4 | 0% | $1.30 | 22m |
All product names, logos, and brands (™/®) are the property of their respective owners; they're used here solely for identification and comparison, and their use does not imply affiliation, endorsement, or sponsorship.