UI-Genie reward model evaluation
Models judge individual Android actions and decide whether a sequence of actions completes a task. The authors report separate F1 scores for both judgments.
Synced covered the 1,050 evaluation pairs. The authors compare six models, including GPT-4o and Gemini 2.5 Pro.
- Made by
- CUHK MMLab and vivo AI Lab
- Released
- May 2025
- Kind
- Phones
- Level
- Everyday
- Scored
- Checked automatically
- Answers
- Short answer
- Measures
- Capability
- Language
- English
- Interaction
- Single turn
- Refresh
- One fixed set
- Run by
- A host
- Last model added
- January 2025
- Scores refreshed
- November 20th, 2026
UI-Genie-RM tops the leaderboard with 80.8% Qwen2.5-VL-72B leads with 66.9%
Best score per entry from the benchmark's paper, read on November 20th, 2026. A paper’s table doesn’t add newer models. The paper reports no combined F1 score, so this is the plain mean of step reward and outcome reward F1. 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. The best model we track is Qwen2.5-VL-72B, at 66.9%. This view shows only models we track, each at its best.
- 1 UI-Genie-RM 80.8%
- 2 Gemini2.5-pro (snapshot unspecified) 73.6%
- 3 Gemini1.5-pro (snapshot unspecified) 68.3%
- 4 GPT-4o (snapshot unspecified) 67.4%
- 5
Qwen2.5-VL-72B 66.9%
- 6
Qwen2.5-VL-7B 57.8%
Source: The UI-Genie reward model evaluation paper
- 1
Qwen2.5-VL-72B 66.9%
- 2
Qwen2.5-VL-7B 57.8%
Source: The UI-Genie reward model evaluation paper
| # | Entry | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | UI-Genie-RM | 80.8% | Official leaderboard | ||
| 2 | Gemini2.5-pro (snapshot unspecified) | 73.6% | Official leaderboard | ||
| 3 | Gemini1.5-pro (snapshot unspecified) | 68.3% | Official leaderboard | ||
| 4 | GPT-4o (snapshot unspecified) | 67.4% | Official leaderboard | ||
| 5 | | 66.9% | Jan 2025 | Official leaderboard | |
| 6 | | 57.8% | Jan 2025 | Official leaderboard |
| # | Model | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | | 66.9% | Jan 2025 | Official leaderboard | |
| 2 | | 57.8% | Jan 2025 | Official 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 UI-Genie reward model evaluation 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
8 of 10 checks pass, 2 warnings. Look first at the Task access check. We don’t know whether the tasks are public. Every benchmark
Catalog record 1 warning 1 pass
- Warning Task access We don’t know whether the tasks are public.
- Pass Confidence Checked against a primary source.
Scores 1 warning 3 pass 1 note
- Pass Independent runs Scores from the benchmark’s paper (6 entries).
- Pass Models scored 2 models we track, plus 4 other entries on its leaderboard.
- Warning Newest score The newest came 22 months ago, in January 2025, for Qwen2.5-VL-72B.
- Note Frontier models None of its models are on the capabilities index.
- Pass Headroom The best is 66.9%, under 85%, so it still separates models.
Leaderboard import 3 pass
- Pass Importer We read the board on arxiv.org: 2 rows for models we track and 4 other entries 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.