Benchmark
Arena Agent Tool Hallucination
Agentic benchmarks measure how well models carry out tasks that take several steps. Higher scores are better.
All agentic benchmarksScores from LMArena data.
- Models measured
- 62
- Scores recorded
- 68
- Sources
- 1
- IQ weight
- 3.08
Score distribution
- -0.002
- Mean
- 0.008
- Median
- 0.022
- Std dev
- -0.035
- 10th percentile
- 0.012
- 90th percentile
- -0.0923
- Min
This benchmark has a weight of 3.08 for chat models, measured over 62 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.
Leaderboard
| # | Model | Score |
|---|---|---|
| 51 | Nemotron 3 Ultra· LMArena data | -0.00422 |
| 52 | Qwen3.7 Max· LMArena data | -0.00459 |
| 53 | Claude Sonnet 5.5High· LMArena data | -0.00597 |
| 54 | Qwen3.8 Max· LMArena data | -0.00954 |
| 55 | Qwen 3.8 Flash Next· LMArena data | -0.0107 |
| 56 | Qwen3.7 Plus· LMArena data | -0.0128 |
| 57 | Claude Haiku 5.5High· LMArena data | -0.0154 |
| 58 | Gemini 3.1 Pro Preview· LMArena data | -0.0174 |
| 59 | Solar Pro 4· LMArena data | -0.0294 |
| 60 | Inkling Small· LMArena data | -0.0295 |
| 61 | Inkling FP4· LMArena data | -0.0354 |
| 62 | Step 5 Preview· LMArena data | -0.0376 |
| 63 | Spark-X2.6· LMArena data | -0.0404 |
| 64 | Hy3 (Deep Infra)· LMArena data | -0.0438 |
| 65 | Mistral Medium 3.5· LMArena data | -0.0538 |
| 66 | MiMo V2.6 Flash· LMArena data | -0.0547 |
| 67 | Beam 1.0 501B A23B· LMArena data | -0.064 |
| 68 | NVIDIA Nemotron 3.5 Super VL 120BNVFP4· LMArena data | -0.0923 |
Showing 51–68 of 68 results.
Data credits: LMArena, CC BY 4.0 (licence)