Benchmark summary
The model builds a Minesweeper-like mine field puzzle in SC# as an agent: it writes the app, compiles it, runs its own tests and fixes the code. Hidden tests check the rules from the task; each hidden test is its own result.
Tests with multiple category tags are counted in each matching category, so category totals may exceed the report total.
| Benchmark items | |||
|---|---|---|---|
| Script / test | Status | Tokens | |
Building Games in SC#: Minesweeper-like mine field puzzle
15 passed1 failed |
93.8% |
IN1 004 606 OUT10 453 | |
| ✓ | MinesweeperThe model builds a keyboard Minesweeper game from the task as an agent; this item passes when the app compiles and can be exported, and each hidden test is its own item.1 AI query
27 tool calls |
Passed | IN1 004 606 OUT10 453 |
| ✓ | Hidden_ChordSpace on a number with exactly that many flagged neighbors reveals its other neighbors; with fewer flags nothing happens.0 AI queries
0 tool callsHiddenTest |
Passed | IN0 OUT0 |
| × | Hidden_ChordWithAWrongFlagA chord with a wrong flag reveals a mine and loses the game; the neighbors are revealed in row order up to that mine.0 AI queries
0 tool callsHiddenTest Assertion expression evaluated to false: cellState[6 * 16 + 4] == 1 && cellState[6 * 16 + 6] == 0
Context: the neighbors before the mine are revealed, the one after it is not (line 827) |
Failed | IN0 OUT0 |
| ✓ | Hidden_CursorMovesEach arrow press moves the cursor one cell, and the cursor stops at the edges.0 AI queries
0 tool callsHiddenTest |
Passed | IN0 OUT0 |
| ✓ | Hidden_FirstRevealIsSafeThe first reveal places exactly 40 mines, none on the revealed cell or its neighbors, so it opens an empty area.0 AI queries
0 tool callsHiddenTest |
Passed | IN0 OUT0 |
| ✓ | Hidden_FlagsF flags and unflags a hidden cell, does nothing on a revealed cell, and the counter shows 40 minus the flags, also below 0.0 AI queries
0 tool callsHiddenTest |
Passed | IN0 OUT0 |
| ✓ | Hidden_FloodRevealAn empty cell opens the whole empty area up to the numbers around it, exactly by the reveal rule.0 AI queries
0 tool callsHiddenTest |
Passed | IN0 OUT0 |
| ✓ | Hidden_FloodStopsAtFlagsFlagged cells are not opened by the flood, and they do not open anything.0 AI queries
0 tool callsHiddenTest |
Passed | IN0 OUT0 |
| ✓ | Hidden_InitialStateAfter AppInit every cell is hidden, there are no mines yet, and the cursor, flags and timer have their starting values.0 AI queries
0 tool callsHiddenTest |
Passed | IN0 OUT0 |
| ✓ | Hidden_LosingRevealing a mine loses the game: every mine is shown, the timer stops, and keys do nothing until Enter.0 AI queries
0 tool callsHiddenTest |
Passed | IN0 OUT0 |
| ✓ | Hidden_RestartEnter starts a new game at any time; that frame does nothing else.0 AI queries
0 tool callsHiddenTest |
Passed | IN0 OUT0 |
| ✓ | Hidden_RevealANumberRevealing a cell with a number opens only that cell; mines are not moved when a test placed them.0 AI queries
0 tool callsHiddenTest |
Passed | IN0 OUT0 |
| ✓ | Hidden_ScreenAfterALossAfter a loss the screen shows the mines and the exploded one in its own color, and no cursor.0 AI queries
0 tool callsHiddenTest |
Passed | IN0 OUT0 |
| ✓ | Hidden_ScreenShowsTheBoardThe screen shows hidden cells, revealed cells, a flag and the cursor at their places.0 AI queries
0 tool callsHiddenTest |
Passed | IN0 OUT0 |
| ✓ | Hidden_TimerThe timer starts with the first reveal and counts frames; the screen shows whole seconds.0 AI queries
0 tool callsHiddenTest |
Passed | IN0 OUT0 |
| ✓ | Hidden_WinningRevealing the last cell without a mine wins: the mines get flags, the counter shows 0 and the timer stops.0 AI queries
0 tool callsHiddenTest |
Passed | IN0 OUT0 |