MUA-RL-8B: benchmark results
Provider: Other.
Aggregate Rating 1553 ± 50, rank #1662 of 10916 rated models, from 24 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| RobustBench-TC - Malformed Response | 67.8 | Tool-call accuracy (%) under the Malform perturbation (trans | 94.1 |
| RobustBench-TC - Schema Drift | 48.2 | Tool-call accuracy (%) under the SchemaD perturbation (trans | 94.1 |
| RobustBench-TC - Server Error | 53.8 | Tool-call accuracy (%) under the 5xxErr perturbation (transi | 82.4 |
| RobustBench-TC - Response-Time Description (Abbreviated Name) | 17.4 | Tool-call accuracy (%) under the TimeDesc-Abbr perturbation | 79.4 |
| RobustBench-TC - Response-Time Description (Neutral Suffix) | 45.1 | Tool-call accuracy (%) under the TimeDesc-N perturbation (re | 76.5 |
| RobustBench-TC - Misleading Description (Neutral Suffix) | 45.1 | Tool-call accuracy (%) under the MisDesc-N perturbation (rew | 70.6 |
| When Simulation Lies | 47.93 | Pert Acc (imported) | 70.6 |
| RobustBench-TC - Rate Limit | 26.1 | Tool-call accuracy (%) under the RateLim perturbation (trans | 64.7 |
| RobustBench-TC - Timeout | 58.8 | Tool-call accuracy (%) under the Timeout perturbation (trans | 64.7 |
| RobustBench-TC - Tool Description Paraphrase | 69.3 | Tool-call accuracy (%) under the ToolPara perturbation (obse | 64.7 |
| RobustBench-TC - Typo | 60.3 | Tool-call accuracy (%) under the Typo perturbation (observat | 64.7 |
| RobustBench-TC - Duplicate Tool With Description | 51.2 | Tool-call accuracy (%) under the Dup-Desc perturbation (acti | 58.8 |
Interactive version: theaggregate.ai/model?slug=mua-rl-8b · How It Works · Data refreshed daily, snapshot 2026-11-08.