agents benchmark
AndroidWorld Leaderboard
AndroidWorld evaluates an agent's ability to operate in real Android GUI environments, completing multi-step tasks by perceiving screen content and executing touch/type actions.
Updated Aug 17, 2026
Models5
Model coverage5
MetricScore
EvidenceB
AndroidWorld Ranking
Higher score ranks better on this benchmark.
Sort by
Rank | Model | Score | Percentile | Participants | Evidence | Evaluated |
|---|
| Rank01 | ModelAC | Score85.3% | Percentile100.0% | Participants5 | EvidenceC | Evaluated |
| Rank02 | ModelAC | Score81.9% | Percentile75.0% | Participants5 | EvidenceC | Evaluated |
| Rank03 | ModelAC | Score81.0% | Percentile50.0% | Participants5 | EvidenceC | Evaluated |
| Rank04 | ModelZA | Score75.7% | Percentile25.0% | Participants5 | EvidenceC | Evaluated |
| Rank05 | ModelAC | Score70.3% | Percentile0.0% | Participants5 | EvidenceC | Evaluated |
AndroidWorld Highlights
The leading models and scores on this benchmark.
AndroidWorld Score Distribution
A closer view of the leading scores on this benchmark.
The Top AI Models for AndroidWorld
The first five results on this benchmark, with official price and output speed added where the model identity can be matched.
Ranking basisThis androidworld AI model leaderboard uses descending score in the benchmark's original unit. The leaderboard ranking keeps matched price and speed data separate from benchmark evidence.
- 01ACQwen3.8 MaxAlibaba Cloud / Qwen Team
- Score
- 85.3%
- Price
- $2.0 input / $6.0 output per 1M tokens
Strengths
- Ranks #1 of 5 compared models
- 100th percentile on this benchmark
- C evidence result
Considerations
- This result measures AndroidWorld, not total model capability
- 02ACQwen3.8-27BAlibaba Cloud / Qwen Team
- Score
- 81.9%
Strengths
- Ranks #2 of 5 compared models
- 75th percentile on this benchmark
- C evidence result
Considerations
- This result measures AndroidWorld, not total model capability
- 03ACQwen3.7-PlusAlibaba Cloud / Qwen Team
- Score
- 81.0%
- Price
- $0.50 input / $3.0 output per 1M tokens
Strengths
- Ranks #3 of 5 compared models
- 50th percentile on this benchmark
- C evidence result
Considerations
- This result measures AndroidWorld, not total model capability
- 04ZAGLM-5V-TurboZhipu AI
- Score
- 75.7%
- Price
- $5.0 input / $22 output per 1M tokens
Strengths
- Ranks #4 of 5 compared models
- 25th percentile on this benchmark
- C evidence result
Considerations
- This result measures AndroidWorld, not total model capability
- 05ACQwen3.6-27BAlibaba Cloud / Qwen Team
- Score
- 70.3%
- Price
- $0.60 input / $3.6 output per 1M tokens
- Speed
- Up to 6.1 tok/s via Novita
Strengths
- Ranks #5 of 5 compared models
- 0th percentile on this benchmark
- C evidence result
Considerations
- This result measures AndroidWorld, not total model capability
Selection summary
Best AI Models for AndroidWorld
Qwen3.8 Max currently leads AndroidWorld with 85.3%. 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.
What is AndroidWorld?
What AndroidWorld measures and how its scores work.
AndroidWorld evaluates an agent's ability to operate in real Android GUI environments, completing multi-step tasks by perceiving screen content and executing touch/type actions.
Scores are shown in ratio. This benchmark is not independently verified and has an evidence level of B.
- Family
- AndroidWorld
- Modality
- multimodal
- Primary category
- agents
- Score direction
- higher
- LLMBoard eligible
- Yes
- Evaluation key
- androidworld|llm-stats-current
Benchmark scores retain their original unit. Overall score eligibility is shown separately.
FAQ
Common questions about AndroidWorld.
Which model scores highest on AndroidWorld?
Qwen3.8 Max is currently ranked first with 85.3%.
What does AndroidWorld measure?
AndroidWorld evaluates an agent's ability to operate in real Android GUI environments, completing multi-step tasks by perceiving screen content and executing touch/type actions.
Is a higher score better?
Yes. Higher values rank better for this benchmark.
How many models are compared?
5 model results are currently shown.
Does this benchmark affect the overall score?
Yes. This benchmark can contribute to the current LLMBoard capability score.