Index · Terminal-Bench Core 0.1.1
count-dataset-tokens
Solution for count-dataset-tokens, task of the Terminal-Bench Core 0.1.1 benchmark.
| Benchmark | Terminal-Bench Core 0.1.1 |
|---|---|
| Task | count-dataset-tokens |
| Category | Agentic terminal use |
| Status | Available — client identification required |
Related identifiers: count dataset tokens.
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=terminal-bench-core-0-1-1&task_id=count-dataset-tokens&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.