Android World Benchmark116 real-world Android tasks

New Benchmark Run

97.4% success rate on Android World Benchmark

FinalRun using Gemini 3 flash and Gemini 3 flash lite completed 113 of 116 Android tasks. The latest run leaves only 3 failures and puts FinalRun at the top of the current leaderboard.

Success Rate

97.4%

Tasks Passed

113

Tasks Failed

3

Tasks Evaluated

116

Leaderboard

Leading the current Android World benchmark

FinalRun now leads a tightly packed field of mobile agents. The leaderboard below reflects the exact ranking and scores from the latest benchmark snapshot.

Rank #1

97.4%

FinalRun (Gemini 3 flash, Gemini 3 flash lite)

113 successful tasks, 3 failures, and a clear lead over the next-best published result.

Android World Benchmark Leaderboard

Current ranking by task success rate

6 entrants
1

FinalRun

Gemini 3 flash, Gemini 3 flash lite

97.4%

2

AutoDevice

gemini 3 pro, sonnet 4.5

94.8%

3

DroidRun

GPT5, Gemini 2.5 Pro

91.4%

4

mobile-use

Gemini 2.5 pro, GPT-5 nano

91.4%

5

Surfer 2

GPT-o3, holo1.5-72b

87.1%

6

gbox ai

Sonnet 4.5

86.2%

Complete Results

Every task from the benchmark run

Filter the full benchmark set to inspect all tasks, only the successful runs, or the three remaining failures.

All 116 Tasks

113 passed, 3 failed, 116 total tasks, 97.4% success rate.

AudioRecorderRecordAudio
AudioRecorderRecordAudioWithFileName
BrowserDraw
BrowserMaze
CameraTakePhoto
CameraTakeVideo
ClockStopWatchPausedVerify
ClockStopWatchRunning
ClockTimerEntry
ContactsAddContact
ContactsNewContactDraft
ExpenseAddMultiple
ExpenseAddSingle
ExpenseDeleteMultiple
ExpenseDeleteMultiple2
ExpenseDeleteSingle
FilesDeleteFile
FilesMoveFile
MarkorChangeNoteContent
MarkorCreateFolder
MarkorCreateNote
MarkorCreateNoteAndSms
MarkorDeleteAllNotes
MarkorDeleteNewestNote
MarkorDeleteNote
MarkorMoveNote
MarkorTranscribeReceipt
NotesIsTodo
NotesMeetingAttendeeCount
NotesRecipeIngredientCount
NotesTodoItemCount
OpenAppTaskEval
OsmAndFavorite
RecipeAddMultipleRecipes
RecipeAddMultipleRecipesFromImage
RecipeAddMultipleRecipesFromMarkor
RecipeAddSingleRecipe
RecipeDeleteDuplicateRecipes
RecipeDeleteMultipleRecipes
RecipeDeleteMultipleRecipesWithNoise
RecipeDeleteSingleRecipe
RecipeDeleteSingleWithRecipeWithNoise
RetroCreatePlaylist
RetroPlayingQueue
RetroSavePlaylist
SaveCopyOfReceiptTaskEval
SimpleCalendarAddOneEvent
SimpleCalendarAddOneEventInTwoWeeks
SimpleCalendarAddOneEventRelativeDay
SimpleCalendarAddOneEventTomorrow
SimpleCalendarAddRepeatingEvent
SimpleCalendarAnyEventsOnDate
SimpleCalendarDeleteEvents
SimpleCalendarDeleteOneEvent
SimpleCalendarEventOnDateAtTime
SimpleCalendarEventsOnDate
SimpleCalendarFirstEventAfterStartTime
SimpleCalendarLocationOfEvent
SimpleCalendarNextEvent
SimpleCalendarNextMeetingWithPerson
SimpleDrawProCreateDrawing
SimpleSmsReply
SimpleSmsReplyMostRecent
SimpleSmsResend
SimpleSmsSend
SimpleSmsSendClipboardContent
SimpleSmsSendReceivedAddress
SportsTrackerActivitiesCountForWeek
SportsTrackerActivitiesOnDate
SportsTrackerActivityDuration
SportsTrackerLongestDistanceActivity
SportsTrackerTotalDistanceForCategoryOverInterval
SportsTrackerTotalDurationForCategoryThisWeek
SystemBluetoothTurnOff
SystemBluetoothTurnOffVerify
SystemBluetoothTurnOn
SystemBluetoothTurnOnVerify
SystemBrightnessMin
SystemBrightnessMinVerify
SystemCopyToClipboard
SystemWifiTurnOff
SystemWifiTurnOffVerify
SystemWifiTurnOn
SystemWifiTurnOnVerify
TasksDueOnDate
TasksHighPriorityTasksDueOnDate
TasksIncompleteTasksOnDate
TurnOffWifiAndTurnOnBluetooth
TurnOnWifiAndOpenApp
BrowserMultiply
ExpenseAddMultipleFromGallery
ExpenseAddMultipleFromMarkor
ExpenseDeleteDuplicates
ExpenseDeleteDuplicates2
MarkorAddNoteHeader
MarkorCreateNoteFromClipboard
MarkorEditNote
MarkorMergeNotes
MarkorTranscribeVideo
OsmAndMarker
RecipeAddMultipleRecipesFromMarkor2
RecipeDeleteDuplicateRecipes2
RecipeDeleteMultipleRecipesWithConstraint
SimpleCalendarDeleteEventsOnRelativeDay
SimpleCalendarEventsInNextWeek
SimpleCalendarEventsInTimeRange
SystemBrightnessMax
SystemBrightnessMaxVerify
TasksCompletedTasksForDate
TasksDueNextWeek
TasksHighPriorityTasks
VlcCreatePlaylist
VlcCreateTwoPlaylists
OsmAndTrack
RecipeDeleteDuplicateRecipes3
RetroPlaylistDuration