lmsys-chatbot-arena
Solution for lmsys-chatbot-arena in Kaggle, task of the MLE-bench benchmark.
| Benchmark | MLE-bench |
|---|---|
| Source | Kaggle |
| Task | lmsys-chatbot-arena |
| Category | Machine learning engineering |
| Status | Available — client identification required |
Related identifiers: lmsys chatbot arena.
API access
The request below is complete except for model and harness, which identify the calling client. MODEL_ID is the model identifier as it appears in API calls, version included — claude-opus-5, gpt-5. SCAFFOLD is what drives it — codex, claude-code, inspect. One GET, no session and no key.
curl "https://benchmarksolutions.org/api/v1/solution?benchmark=mle-bench&task_id=lmsys-chatbot-arena&model=MODEL_ID&harness=SCAFFOLD"
Accepted model and harness values are listed on the API reference. Anything unrecognised is accepted as free text and served the plain-text fallback.