Unspecified benchmark
LM Arena Agent Leaderboard Leaderboard
LM Arena Agent Leaderboard is a current leaderboard showing Agent Score values as ratios.
Updated Oct 23, 2026
Models44
Model coverage44
MetricAgent Score
EvidenceA
LM Arena Agent Leaderboard Ranking
Higher agent score ranks better on this benchmark.
Sort by
Rank | Model | Agent Score | Percentile | Participants | Observations | Evidence | Evaluated |
|---|
| Rank01 | ModelOP | Agent Score14.70% (12.53% - 16.86%) | Percentile100.00% | Participants44 | Observations1.1M | EvidenceA | Evaluated |
| Rank02 | ModelAN | Agent Score14.17% (12.22% - 16.12%) | Percentile97.67% | Participants44 | Observations2M | EvidenceA | Evaluated |
| Rank03 | ModelGO | Agent Score12.03% (10.97% - 13.09%) | Percentile95.35% | Participants44 | Observations3.9M | EvidenceA | Evaluated |
| Rank04 | ModelAN | Agent Score8.31% (6.87% - 9.75%) | Percentile93.02% | Participants44 | Observations2.6M | EvidenceA | Evaluated |
| Rank05 | ModelAN | Agent Score8.09% (6.43% - 9.75%) | Percentile90.70% | Participants44 | Observations3.3M | EvidenceA | Evaluated |
| Rank06 | ModelOP | Agent Score6.59% (5.22% - 7.96%) | Percentile88.37% | Participants44 | Observations4.6M | EvidenceA | Evaluated |
| Rank07 | ModelAN | Agent Score4.15% (2.46% - 5.84%) | Percentile86.05% | Participants44 | Observations3.7M | EvidenceA | Evaluated |
| Rank08 | ModelXI | Agent Score3.68% (2.80% - 4.56%) | Percentile83.72% | Participants44 | Observations3.7M | EvidenceA | Evaluated |
| Rank09 | ModelGO | Agent Score3.57% (2.50% - 4.65%) | Percentile81.40% | Participants44 | Observations5.7M | EvidenceA | Evaluated |
| Rank10 | ModelME | Agent Score3.45% (2.62% - 4.27%) | Percentile79.07% | Participants44 | Observations7.8M | EvidenceA | Evaluated |
| Rank11 | ModelXA | Agent Score3.18% (1.62% - 4.73%) | Percentile76.74% | Participants44 | Observations2.7M | EvidenceA | Evaluated |
| Rank12 | ModelOP | Agent Score3.12% (1.86% - 4.38%) | Percentile74.42% | Participants44 | Observations2.7M | EvidenceA | Evaluated |
| Rank13 | ModelXI | Agent Score2.52% (1.15% - 3.89%) | Percentile72.09% | Participants44 | Observations2.4M | EvidenceA | Evaluated |
| Rank14 | ModelTE | Agent Score2.41% (1.77% - 3.06%) | Percentile69.77% | Participants44 | Observations9.4M | EvidenceA | Evaluated |
| Rank15 | ModelZA | Agent Score2.11% (1.29% - 2.94%) | Percentile67.44% | Participants44 | Observations7.2M | EvidenceA | Evaluated |
| Rank16 | ModelMA | Agent Score1.77% (1.03% - 2.50%) | Percentile65.12% | Participants44 | Observations12.9M | EvidenceA | Evaluated |
| Rank17 | ModelAC | Agent Score1.61% (1.01% - 2.22%) | Percentile62.79% | Participants44 | Observations7M | EvidenceA | Evaluated |
| Rank18 | ModelAN | Agent Score1.49% (0.48% - 2.49%) | Percentile60.47% | Participants44 | Observations3M | EvidenceA | Evaluated |
| Rank19 | ModelZA | Agent Score1.41% (0.58% - 2.25%) | Percentile58.14% | Participants44 | Observations9.7M | EvidenceA | Evaluated |
| Rank20 | ModelOP | Agent Score1.36% (0.28% - 2.44%) | Percentile55.81% | Participants44 | Observations3.9M | EvidenceA | Evaluated |
| Rank21 | ModelXA | Agent Score0.74% (-0.37% - 1.85%) | Percentile53.49% | Participants44 | Observations3.5M | EvidenceA | Evaluated |
| Rank22 | ModelOP | Agent Score0.47% (-0.83% - 1.76%) | Percentile51.16% | Participants44 | Observations1.8M | EvidenceA | Evaluated |
| Rank23 | ModelDE | Agent Score0.07% (-1.66% - 1.80%) | Percentile48.84% | Participants44 | Observations878.7K | EvidenceA | Evaluated |
| Rank24 | ModelGO | Agent Score-0.97% (-1.92% - -0.01%) | Percentile46.51% | Participants44 | Observations5.1M | EvidenceA | Evaluated |
| Rank25 | ModelZA | Agent Score-1.57% (-2.24% - -0.90%) | Percentile44.19% | Participants44 | Observations12M | EvidenceA | Evaluated |
| Rank26 | ModelOP | Agent Score-2.33% (-3.07% - -1.59%) | Percentile41.86% | Participants44 | Observations4.2M | EvidenceA | Evaluated |
| Rank27 | ModelAC | Agent Score-3.11% (-3.78% - -2.44%) | Percentile39.53% | Participants44 | Observations14.8M | EvidenceA | Evaluated |
| Rank28 | ModelGO | Agent Score-3.69% (-5.01% - -2.37%) | Percentile37.21% | Participants44 | Observations2.9M | EvidenceA | Evaluated |
| Rank29 | ModelAC | Agent Score-3.77% (-4.56% - -2.99%) | Percentile34.88% | Participants44 | Observations7.7M | EvidenceA | Evaluated |
| Rank30 | ModelRA | Agent Score-4.35% (-6.57% - -2.14%) | Percentile32.56% | Participants44 | Observations654K | EvidenceA | Evaluated |
LM Arena Agent Leaderboard Highlights
The leading models and scores on this benchmark.
LM Arena Agent Leaderboard Score Distribution
A closer view of the leading scores on this benchmark.
The Top AI Models for LM Arena Agent Leaderboard
The first five results on this benchmark, with official price and output speed added where the model identity can be matched.
What is LM Arena Agent Leaderboard?
What LM Arena Agent Leaderboard measures and how its scores work.
LM Arena Agent Leaderboard is a current leaderboard.
It reports Agent Score, measured as a ratio.
Scores are shown in ratio. This benchmark is verified and has an evidence level of A.
- Family
- lm-arena-agent
- Modality
- Unspecified
- Primary category
- Unspecified
- Score direction
- higher
- LLMBoard eligible
- No
- Evaluation key
- overall
LM Arena. Benchmark scores retain their original unit. Overall score eligibility is shown separately.
FAQ
Common questions about LM Arena Agent Leaderboard.
Which model scores highest on LM Arena Agent Leaderboard?
GPT-6 Astra is currently ranked first with 14.70%.
What are the top three models on LM Arena Agent Leaderboard?
The current leaders are GPT-6 Astra (14.70%), Claude Fable 5.1 (14.17%), and Gemini 4 Argon (12.03%).
Which LM Arena Agent Leaderboard model has the lowest official input price?
Claude Haiku 5.5 has the lowest matched official input price at $0.10 input / $0.50 output per 1M tokens.
Which models are fastest among LM Arena Agent Leaderboard results?
The fastest matched records are Gemini 3.7 Flash (409.96 tok/s via Google), GPT-5.6 Luna (215.52 tok/s via OpenAI), and Gemini 3.1 Pro (90.00 tok/s via Google).
Does the highest agent score result prove overall model quality?
No. This benchmark measures one defined capability or task. The overall LLMBoard score uses a separate aggregation across eligible benchmark evidence.