multimodal benchmark
AndroidWorld_SR
AndroidWorld Success Rate (SR) benchmark - A dynamic benchmarking environment for autonomous agents operating on Android devices. Evaluates agents on 116 programmatic tasks across 20 real-world Android apps using multimodal inputs (screen screenshots, accessibility trees, and natural language instructions). Measures success rate of agents completing tasks like sending messages, creating calendar events, and navigating mobile interfaces. Published at ICLR 2025. Best current performance: 30.6% success rate (M3A agent) vs 80.0% human performance.
Updated Aug 7, 2026
Published models8
Registry coverage8
MetricScore
EvidenceB
AndroidWorld_SR leaderboard
Sorted by the source-provided rank. Higher score is better according to the registry.
| 1 | AC | 71.1% | 100.0% | 8 | C | |
| 2 | AC | 66.4% | 85.7% | 8 | C | |
| 3 | AC | 64.2% | 71.4% | 8 | C | |
| 4 | AC | 63.7% | 57.1% | 8 | C | |
| 5 | AC | 63.7% | 42.9% | 8 | C | |
| 6 | AC | 35.0% | 28.6% | 8 | C | |
| 7 | AC | 25.5% | 14.3% | 8 | C | |
| 8 | AC | 22.0% | 0.0% | 8 | C |
Score distribution
Top published rows on the benchmark's original scale.
AndroidWorld_SR highlights
The top published results on this benchmark's own scale.
What is AndroidWorld_SR?
Definition and scoring fields from the benchmark registry.
AndroidWorld Success Rate (SR) benchmark - A dynamic benchmarking environment for autonomous agents operating on Android devices. Evaluates agents on 116 programmatic tasks across 20 real-world Android apps using multimodal inputs (screen screenshots, accessibility trees, and natural language instructions). Measures success rate of agents completing tasks like sending messages, creating calendar events, and navigating mobile interfaces. Published at ICLR 2025. Best current performance: 30.6% success rate (M3A agent) vs 80.0% human performance.
Scores are shown in ratio. The current registry marks this benchmark as not independently verified with evidence level B.
- Family
- AndroidWorld_SR
- Modality
- multimodal
- Primary category
- multimodal
- Score direction
- higher
- LLMBoard eligible
- No
- Evaluation key
- androidworld-sr|llm-stats-current
Source-native results are preserved. Eligibility for the overall LLMBoard score is a separate policy decision.
FAQ
Common questions about AndroidWorld_SR.
Which model scores highest on AndroidWorld_SR?
Qwen3.5-35B-A3B is currently ranked first with 71.1%.
What does AndroidWorld_SR measure?
AndroidWorld Success Rate (SR) benchmark - A dynamic benchmarking environment for autonomous agents operating on Android devices. Evaluates agents on 116 programmatic tasks across 20 real-world Android apps using multimodal inputs (screen screenshots, accessibility trees, and natural language instructions). Measures success rate of agents completing tasks like sending messages, creating calendar events, and navigating mobile interfaces. Published at ICLR 2025. Best current performance: 30.6% success rate (M3A agent) vs 80.0% human performance.
Is a higher score better?
Yes. Higher values rank better for this benchmark.
How many models are compared?
8 unique published model results are currently shown.
Does this benchmark affect the overall score?
This benchmark is preserved as source-native evidence but is not eligible for the current overall score.