Unspecified benchmark
LM Arena Agent Tool Hallucination is a current leaderboard reporting Agent Score as a ratio.
Updated Sep 8, 2026
Higher agent score ranks better on this benchmark.
Rank | Model | Agent Score | Percentile | Participants | Observations | Evidence | Evaluated |
|---|
| Rank01 | ModelOP | Agent Score0.78% (0.65% - 0.91%) | Percentile100.00% | Participants49 | Observations3M | EvidenceA | Evaluated |
| Rank02 | ModelOP | Agent Score0.78% (0.65% - 0.91%) | Percentile97.92% | Participants49 | Observations981.2K | EvidenceA | Evaluated |
| Rank03 | ModelXA | Agent Score0.78% (0.65% - 0.91%) | Percentile95.83% | Participants49 | Observations1.8M | EvidenceA | Evaluated |
| Rank04 | ModelDE | Agent Score0.78% (0.65% - 0.91%) | Percentile93.75% | Participants49 | Observations2.7M | EvidenceA | Evaluated |
| Rank05 | ModelOP | Agent Score0.78% (0.65% - 0.91%) | Percentile91.67% | Participants49 | Observations1.5M | EvidenceA | Evaluated |
| Rank06 | ModelXA | Agent Score0.78% (0.65% - 0.91%) | Percentile89.58% | Participants49 | Observations1.6M | EvidenceA | Evaluated |
| Rank07 | ModelOP | Agent Score0.78% (0.65% - 0.91%) | Percentile87.50% | Participants49 | Observations1.8M | EvidenceA | Evaluated |
| Rank08 | ModelMA | Agent Score0.78% (0.65% - 0.91%) | Percentile85.42% | Participants49 | Observations7.8M | EvidenceA | Evaluated |
| Rank09 | ModelZA | Agent Score0.78% (0.65% - 0.91%) | Percentile83.33% | Participants49 | Observations3.7M | EvidenceA | Evaluated |
| Rank10 | ModelZA | Agent Score0.78% (0.65% - 0.91%) | Percentile81.25% | Participants49 | Observations1.8M | EvidenceA | Evaluated |
| Rank11 | ModelMA | Agent Score0.78% (0.65% - 0.91%) | Percentile79.17% | Participants49 | Observations304.3K | EvidenceA | Evaluated |
| Rank12 | ModelMA | Agent Score0.78% (0.65% - 0.91%) | Percentile77.08% | Participants49 | Observations383.4K | EvidenceA | Evaluated |
| Rank13 | ModelAN | Agent Score0.78% (0.65% - 0.91%) | Percentile75.00% | Participants49 | Observations488.8K | EvidenceA | Evaluated |
| Rank14 | ModelZA | Agent Score0.78% (0.65% - 0.91%) | Percentile72.92% | Participants49 | Observations4M | EvidenceA | Evaluated |
| Rank15 | ModelOP | Agent Score0.78% (0.65% - 0.91%) | Percentile70.83% | Participants49 | Observations3.3M | EvidenceA | Evaluated |
| Rank16 | ModelAN | Agent Score0.78% (0.65% - 0.90%) | Percentile68.75% | Participants49 | Observations1.2M | EvidenceA | Evaluated |
| Rank17 | ModelAN | Agent Score0.78% (0.65% - 0.90%) | Percentile66.67% | Participants49 | Observations1.7M | EvidenceA | Evaluated |
| Rank18 | ModelME | Agent Score0.77% (0.65% - 0.90%) | Percentile64.58% | Participants49 | Observations798.1K | EvidenceA | Evaluated |
| Rank19 | ModelME | Agent Score0.77% (0.64% - 0.89%) | Percentile62.50% | Participants49 | Observations3.2M | EvidenceA | Evaluated |
| Rank20 | ModelAN | Agent Score0.76% (0.63% - 0.88%) | Percentile60.42% | Participants49 | Observations2.2M | EvidenceA | Evaluated |
| Rank21 | ModelDE | Agent Score0.76% (0.63% - 0.88%) | Percentile58.33% | Participants49 | Observations5.4M | EvidenceA | Evaluated |
| Rank22 | ModelAN | Agent Score0.72% (0.59% - 0.86%) | Percentile56.25% | Participants49 | Observations1.4M | EvidenceA | Evaluated |
| Rank23 | ModelGO | Agent Score0.71% (0.58% - 0.85%) | Percentile54.17% | Participants49 | Observations1.6M | EvidenceA | Evaluated |
| Rank24 | ModelAN | Agent Score0.71% (0.55% - 0.87%) | Percentile52.08% | Participants49 | Observations1.3M | EvidenceA | Evaluated |
| Rank25 | ModelGO | Agent Score0.70% (0.54% - 0.85%) | Percentile50.00% | Participants49 | Observations531.5K | EvidenceA | Evaluated |
| Rank26 | ModelGO | Agent Score0.69% (0.51% - 0.87%) | Percentile47.92% | Participants49 | Observations922K | EvidenceA | Evaluated |
| Rank27 | ModelXA | Agent Score0.67% (0.54% - 0.80%) | Percentile45.83% | Participants49 | Observations1.1M | EvidenceA | Evaluated |
| Rank28 | ModelTE | Agent Score0.66% (0.51% - 0.80%) | Percentile43.75% | Participants49 | Observations1.2M | EvidenceA | Evaluated |
| Rank29 | ModelAN | Agent Score0.63% (0.48% - 0.78%) | Percentile41.67% | Participants49 | Observations2.7M | EvidenceA | Evaluated |
| Rank30 | ModelMI | Agent Score0.61% (0.44% - 0.79%) | Percentile39.58% | Participants49 | Observations772.3K | EvidenceA | Evaluated |
The leading models and scores on this benchmark.
A closer view of the leading scores on this benchmark.
The first five results on this benchmark, with official price and output speed added where the model identity can be matched.
What LM Arena Agent Tool Hallucination measures and how its scores work.
A current leaderboard for LM Arena Agent Tool Hallucination.
It reports Agent Score, measured as a ratio.
Scores are shown in ratio. This benchmark is verified and has an evidence level of A.
LM Arena. Benchmark scores retain their original unit. Overall score eligibility is shown separately.
Common questions about LM Arena Agent Tool Hallucination.
GPT-5.6 Sol is currently ranked first with 0.78%.
The current leaders are GPT-5.6 Sol (0.78%), GPT-5.6 Terra (0.78%), and Grok 4.6 (0.78%).
GLM-5.3-Flash has the lowest matched official input price at $0.08 input / $0.25 output per 1M tokens.
The fastest matched records are GLM-5.3 (472.69 tok/s via FriendliAI), Gemini 3.5 Flash (210.94 tok/s via Google), and GPT-5.5 (134.94 tok/s via OpenAI).
No. This benchmark measures one defined capability or task. The overall LLMBoard score uses a separate aggregation across eligible benchmark evidence.
It reports Agent Score, measured as a ratio.
Yes. Higher values rank better for this benchmark.
49 model results are currently shown.
No. This benchmark is shown for reference but does not contribute to the overall score.
Ranking basisThis lm arena agent tool hallucination AI model leaderboard uses descending agent score in the benchmark original unit. The leaderboard ranking keeps matched price and speed data separate from benchmark evidence.
Selection summary
GPT-5.6 Sol currently leads LM Arena Agent Tool Hallucination with 0.78%. It is the top model on this specific benchmark, while the best LLM for the broader task should also be checked against other benchmarks, price, and runtime.
Use this leaderboard with the supporting benchmark results and coverage details above. A leaderboard position summarizes the selected ranking signal; it does not replace workload-specific testing.