GUI Knowledge Bench
Models answer 3,483 questions about interface states, actions, and procedures. They inspect screenshots and task sequences from 292 applications across six platforms.
Alibaba reports 75.5% for GUI-Owl 1.5. The makers’ online leaderboard ranks nine models by the unweighted mean of eight scores.
- Released
- October 2025
- Kind
- Finding things on screen
- Level
- Everyday
- Scored
- Checked automatically
- Answers
- Multiple choice
- Tasks
- All public
- Measures
- Capability
- Language
- English
- Interaction
- Single turn
- Items from
- Real work and use
- Refresh
- One fixed set
- Run by
- Anyone
- Strongest model tested
- o3 #84 on the capability index
- Last model added
- August 2025
- Scores refreshed
- November 20th, 2026
o3 leads with 78.9% among models released by August 2025 o3 leads with 78.9% among models released by August 2025
Best score per entry from the benchmark's own leaderboard, read on November 20th, 2026. Where a model ran with several settings or harnesses, we keep its best. Names in gray aren't in our model list: fine-tunes, agent systems, products and older models. They don't count toward leaders. This view shows only models we track, each at its best.
The newest model with a score here came out in August 2025, and nobody has tested a newer one. Among the 20 highest-ranked models on the capability index, these have no score yet:
- GPT-6.2 Astra
- Claude Fable 5.2
-
Gemini 4 Carbon
- GPT-6 Astra
- and 16 more
- 1 o3 78.9%
- 2
Gemini 2.5 Pro Jun 2025 77.1%
- 3 GPT-5 Chat 76.6%
- 4 Doubao-V-Pro 70.1%
- 5
Qwen2.5-VL-72B 68.4%
- 6 Claude Sonnet 4 66.9%
- 7
Qwen2.5-VL-7B 49.3%
- 8
UI-TARS-1.5 48.6%
- 9 GLM-4.5 44.6%
Source: Official GUI Knowledge Bench leaderboard
- 1 o3 78.9%
- 2
Gemini 2.5 Pro Jun 2025 77.1%
- 3 GPT-5 Chat 76.6%
- 4
Qwen2.5-VL-72B 68.4%
- 5 Claude Sonnet 4 66.9%
- 6
Qwen2.5-VL-7B 49.3%
- 7
UI-TARS-1.5 48.6%
- 8 GLM-4.5 44.6%
Source: Official GUI Knowledge Bench leaderboard
| # | Entry | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | o3 | 78.9% | Apr 2025 | Official leaderboard | |
| 2 | | 77.1% | Jun 2025 | Official leaderboard | |
| 3 | GPT-5 Chat | 76.6% | Aug 2025 | Official leaderboard | |
| 4 | Doubao-V-Pro | 70.1% | Official leaderboard | ||
| 5 | | 68.4% | Jan 2025 | Official leaderboard | |
| 6 | Claude Sonnet 4 | 66.9% | May 2025 | Official leaderboard | |
| 7 | | 49.3% | Jan 2025 | Official leaderboard | |
| 8 | | 48.6% | Apr 2025 | Official leaderboard | |
| 9 | GLM-4.5 | 44.6% | Aug 2025 | Official leaderboard |
| # | Model | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | o3 | 78.9% | Apr 2025 | Official leaderboard | |
| 2 | | 77.1% | Jun 2025 | Official leaderboard | |
| 3 | GPT-5 Chat | 76.6% | Aug 2025 | Official leaderboard | |
| 4 | | 68.4% | Jan 2025 | Official leaderboard | |
| 5 | Claude Sonnet 4 | 66.9% | May 2025 | Official leaderboard | |
| 6 | | 49.3% | Jan 2025 | Official leaderboard | |
| 7 | | 48.6% | Apr 2025 | Official leaderboard | |
| 8 | GLM-4.5 | 44.6% | Aug 2025 | Official leaderboard |
Rankings by part
We read 8 parts from the benchmark’s own leaderboard, each ranking models on one part of the test, such as State, Widget and Layout. o3 leads 7 of them, and Gemini 2.5 Pro leads the other one. Arrows show how far a model moved from its overall rank among the models we track.
State
8 models we track have a score here.
- 1 o3 83%
- 2
Gemini 2.5 Pro Jun 2025 81.2%
- 3 GPT-5 Chat 78.9%
- 4 Claude Sonnet 4 ↑1 70.2%
- 5
Qwen2.5-VL-72B ↓1 69.7%
- 6
Qwen2.5-VL-7B 53.2%
- 7
UI-TARS-1.5 49.5%
- 8 GLM-4.5 49.5%
Widget
8 models we track have a score here.
- 1
Gemini 2.5 Pro Jun 2025 ↑1 84.4%
- 2 o3 ↓1 84.1%
- 3 GPT-5 Chat 84.1%
- 4 Claude Sonnet 4 ↑1 78.4%
- 5
Qwen2.5-VL-72B ↓1 77.5%
- 6
Qwen2.5-VL-7B 67.8%
- 7
UI-TARS-1.5 59.5%
- 8 GLM-4.5 48.1%
Layout
8 models we track have a score here.
- 1 o3 88.4%
- 2 GPT-5 Chat ↑1 88.4%
- 3
Gemini 2.5 Pro Jun 2025 ↓1 87.1%
- 4
Qwen2.5-VL-72B 80%
- 5 Claude Sonnet 4 78.1%
- 6
Qwen2.5-VL-7B 60%
- 7
UI-TARS-1.5 59.4%
- 8 GLM-4.5 53.5%
Effect
8 models we track have a score here.
- 1 o3 76.5%
- 2 GPT-5 Chat ↑1 73.1%
- 3
Gemini 2.5 Pro Jun 2025 ↓1 71.7%
- 4
Qwen2.5-VL-72B 62.9%
- 5
Qwen2.5-VL-7B ↑1 52.6%
- 6 Claude Sonnet 4 ↓1 41.5%
- 7 GLM-4.5 ↑1 27.1%
- 8
UI-TARS-1.5 ↓1 22.8%
Type
8 models we track have a score here.
- 1 o3 76%
- 2
Gemini 2.5 Pro Jun 2025 73.3%
- 3 GPT-5 Chat 71.5%
- 4
Qwen2.5-VL-72B 64.9%
- 5 Claude Sonnet 4 62.5%
- 6
UI-TARS-1.5 ↑1 59.1%
- 7
Qwen2.5-VL-7B ↓1 50.6%
- 8 GLM-4.5 17.6%
Parameter
8 models we track have a score here.
- 1 o3 60.3%
- 2
Gemini 2.5 Pro Jun 2025 60.3%
- 3 GPT-5 Chat 57.9%
- 4 Claude Sonnet 4 ↑1 45.8%
- 5
Qwen2.5-VL-72B ↓1 44.9%
- 6
Qwen2.5-VL-7B 44.5%
- 7
UI-TARS-1.5 43.5%
- 8 GLM-4.5 38.1%
Goal
8 models we track have a score here.
- 1 o3 67.6%
- 2 GPT-5 Chat ↑1 67.4%
- 3
Gemini 2.5 Pro Jun 2025 ↓1 66.5%
- 4 Claude Sonnet 4 ↑1 63.8%
- 5
Qwen2.5-VL-72B ↓1 61.9%
- 6
UI-TARS-1.5 ↑1 39.8%
- 7 GLM-4.5 ↑1 30.9%
- 8
Qwen2.5-VL-7B ↓2 17%
Plan
8 models we track have a score here.
- 1 o3 95.5%
- 2 Claude Sonnet 4 ↑3 94.8%
- 3
Gemini 2.5 Pro Jun 2025 ↓1 92.6%
- 4 GLM-4.5 ↑4 91.9%
- 5 GPT-5 Chat ↓2 91.3%
- 6
Qwen2.5-VL-72B ↓2 85.4%
- 7
UI-TARS-1.5 55.3%
- 8
Qwen2.5-VL-7B ↓2 48.9%
Number one over time
Lab claims count from the day the card came out, independent runs from the model’s release.
Source: Official GUI Knowledge Bench leaderboard
Scores over time
All the scores we have, one dot each, so a model tested at several settings appears more than once.
Source: Official GUI Knowledge Bench leaderboard
No lab model card reports it yet
We read the headline results of 540 launch posts from 57 labs, and none of them list this benchmark.
More in Computer and browser use
- BrowseComp In use Hard-to-find facts that need persistent web browsing. Short verifiable answers graded for exact correctness. Claude Fable 5 93.3%
- ScreenSpot-Pro In use GUI grounding on high-resolution professional-software screenshots. The model must locate the target UI element. GPT-6 Astra 92.7%
- AndroidWorld In use Agents complete 116 tasks across 20 real Android apps in an emulator. Each task checks the device's system state to judge success. GPT-6 Astra 100%
- OSWorld 2.1 In use Agents complete 108 long-horizon workflows across 31 self-hosted websites and desktop apps, averaging 318 tool calls. Scored by binary completion and checkpoint partial credit. Claude Opus 5 44.3%
- WebArena-Verified In use Agents complete 812 audited WebArena tasks on self-hosted websites. Deterministic evaluators check the agent's answer and captured network traces.
Muse Spark 1.1 69%
- OSWorld 2.1 (offline subset) In use Agents complete the 82 OSWorld 2.1 tasks on the official offline list, with no internet access. Scored by checkpoint partial credit and strict pass rate. Claude Fable 5.2 89.3%
Data audit
9 of 10 checks pass, 1 warning. Look first at the Newest score check. The newest came 15 months ago, in August 2025, for GPT-5 Chat. Every benchmark
Catalog record 2 pass
- Pass Task access All tasks are public.
- Pass Confidence Checked against a primary source.
Scores 1 warning 3 pass 1 note
- Pass Independent runs Scores from the official leaderboard (9 entries).
- Pass Models scored 8 models we track, plus 1 other entry on its leaderboard.
- Warning Newest score The newest came 15 months ago, in August 2025, for GPT-5 Chat.
- Note Frontier models None of the top 10 has a score. The strongest tested is o3 (#84).
- Pass Headroom The best is 78.9%, under 85%, so it still separates models.
Leaderboard import 3 pass
- Pass Importer We read the board on computer-use-agents.github.io: 8 rows for models we track and 1 other entry at the last read.
- Pass Last read November 20th, 2026.
- Pass Read worked The last read worked.
Consistency 1 pass 1 note
- Note Lab claims No model has both a lab number and an independent run.
- Pass Known issues No published audit found problems.