AgentMonBench
Harbor: pendingA software-engineering benchmark evaluating monitors' ability to identify consequential agent decisions and locate evidence for user oversight across three subsets.
- Samples
- 300
- Released
- 2026
- Scores
- 8
HygieneRoboBench
HygieneRoboBench: Benchmarking Hygiene-Aware Planning for Household Robots
A benchmark with 624 instances across 134 task families evaluating household robots' ability to plan safe, hygiene-aware task continuations considering contact history and user priorities.
How to run
ltb · hygienerobobench
$ npm i -g lettokenburn-cli$ ltb bench info hygienerobobench
External harness. No Harbor adapter has been generated for this benchmark yet. `ltb bench info` prints the known links.
Runs happen on your machine with your own subscription; only the score summary is pushed. CLI docs
Scores
5 sourced entries| Harness | Source | Confidence | Link | |||
|---|---|---|---|---|---|---|
| DeepSeek V4.1 Flash | 90.7 | N/A | Independent | medium | Oct 6, 2026 | arxiv.org |
| Gemini 3.6 Flash | 89.7 | N/A | Independent | medium | Oct 6, 2026 | arxiv.org |
| GPT-5.6 Sol | 78.5 | N/A | Independent | medium | Oct 6, 2026 | arxiv.org |
| MiniMax-M3 | 67.0 | N/A | Independent | medium | Oct 6, 2026 | arxiv.org |
| GPT-5.6 Luna | 46.6 | N/A | Independent | medium | Oct 6, 2026 | arxiv.org |
Every number links to where it was published. Low-confidence rows are muted and need per-cell verification.
Reproduced by the community
All runsLoading community runs…
About
HygieneRoboBench contains 624 instances across 134 task families in seven household areas. It tests planners on contamination propagation, resource trade-offs, and user priorities using controlled history and event comparisons.
Scoring focuses on safe resolution and optimal safe resolution rates. Evaluation of seven planners shows Hygiene-NSP achieves the highest safe resolution (94.4%) and optimal safe resolution (90.4%) rates.