run-0101
capability
citable URL: https://halobench.com/records/run-0101/ — this address never moves; the anchor /records/#run-0101 keeps resolving
date 2026-08-10 · window 2026-08-10T03:06:47Z → 2026-08-10T08:19:49Z
suite tau2-bench-airline@668d3bc · config cfg-0026
depth 32,768 tokens · scoring deterministic
Metrics
| metric | value |
|---|---|
| tasks_passed | 8 |
| tasks_total | 9 |
| mean_score | 0.8889 |
turns: total 228 · median 22 · p90 48 · max 48 — machine-written trailer, parsed at build time
Guard chain
not a performance run — no guard required
Per-task trace — the full record, no truncation
| task | reward | turns | seconds |
|---|---|---|---|
| 0 | 1 | 14 | 97 |
| 1 | 1 | 24 | 158 |
| 2 | 1 | 40 | 363 |
| 3 | 1 | 14 | 67 |
| 4 | 1 | 48 | 252 |
| 5 | 1 | 22 | 182 |
| 6 | 1 | 10 | 59 |
| 7 | null (infrastructure) | 0 | — |
| 8 | 0 | 38 | 288 |
| 9 | 1 | 18 | 183 |
reward null = infrastructure error, excluded from every paired computation
Cited by — computed at build time, never stored
model pages qwen35-122b