multimodal benchmark
ScreenSpot Leaderboard
ScreenSpot is a realistic GUI grounding benchmark covering mobile, desktop, and web environments.
Updated Sep 7, 2026
Models16
Model coverage16
MetricScore
EvidenceB
ScreenSpot Ranking
Higher score ranks better on this benchmark.
Sort by
Rank | Model | Score | Percentile | Participants | Evidence | Evaluated |
|---|
| Rank01 | ModelAC | Score95.80% | Percentile100.00% | Participants16 | EvidenceC | Evaluated |
| Rank02 | ModelAC | Score95.70% | Percentile93.33% | Participants16 | EvidenceC | Evaluated |
| Rank03 | ModelAC | Score95.40% | Percentile86.67% | Participants16 | EvidenceC | Evaluated |
| Rank04 | ModelAC | Score95.40% | Percentile80.00% | Participants16 | EvidenceC | Evaluated |
| Rank05 | ModelAC | Score94.70% | Percentile73.33% | Participants16 | EvidenceC | Evaluated |
| Rank06 | ModelAC | Score94.70% | Percentile66.67% | Participants16 | EvidenceC | Evaluated |
| Rank07 | ModelAC | Score94.40% | Percentile60.00% | Participants16 | EvidenceC | Evaluated |
| Rank08 | ModelAC | Score94.00% | Percentile53.33% | Participants16 | EvidenceC | Evaluated |
| Rank09 | ModelAC | Score93.60% | Percentile46.67% | Participants16 | EvidenceC | Evaluated |
| Rank10 | ModelAC | Score92.90% | Percentile40.00% | Participants16 | EvidenceC | Evaluated |
| Rank11 | ModelAC | Score88.50% | Percentile33.33% | Participants16 | EvidenceC | Evaluated |
| Rank12 | ModelAM Nova 2 ProAmazon | Score88.10% | Percentile26.67% | Participants16 | EvidenceC | Evaluated |
| Rank13 | ModelAC | Score87.10% | Percentile20.00% | Participants16 | EvidenceC | Evaluated |
| Rank14 | ModelAMNova 2 OmniAmazon | Score85.40% | Percentile13.33% | Participants16 | EvidenceC | Evaluated |
| Rank15 | ModelAC | Score84.70% | Percentile6.67% | Participants16 | EvidenceC | Evaluated |
| Rank16 | ModelAMNova 2 LiteAmazon | Score83.30% | Percentile0.00% | Participants16 | EvidenceC | Evaluated |
ScreenSpot Highlights
The leading models and scores on this benchmark.
ScreenSpot Score Distribution
A closer view of the leading scores on this benchmark.
The Top AI Models for ScreenSpot
The first five results on this benchmark, with official price and output speed added where the model identity can be matched.
What is ScreenSpot?
What ScreenSpot measures and how its scores work.
ScreenSpot is a multimodal benchmark with over 1,200 instructions from iOS, Android, macOS, Windows and Web environments, including annotated text and icon/widget element types.
It measures visual GUI agents' ability to locate screen elements accurately from natural language instructions, using the Score metric as a ratio.
Scores are shown in ratio. This benchmark is not independently verified and has an evidence level of B.
- Family
- ScreenSpot
- Modality
- multimodal
- Primary category
- multimodal
- Score direction
- higher
- LLMBoard eligible
- No
- Evaluation key
- overall
LLM Stats. Benchmark scores retain their original unit. Overall score eligibility is shown separately.
FAQ
Common questions about ScreenSpot.
Which model scores highest on ScreenSpot?
Qwen3 VL 32B Instruct is currently ranked first with 95.80%.
What are the top three models on ScreenSpot?
The current leaders are Qwen3 VL 32B Instruct (95.80%), Qwen3 VL 32B Thinking (95.70%), and Qwen3 VL 235B A22B Thinking (95.40%).
Which ScreenSpot model has the lowest official input price?
Nova 2 Lite has the lowest matched official input price at $0.33 input / $2.8 output per 1M tokens.
Which models are fastest among ScreenSpot results?
The fastest matched records are Qwen3 VL 4B Thinking (8.77 tok/s via DeepInfra).
Does the highest 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.