TIS-100 Bench: Wave Collapse Supervisor

Accumulate four input streams independently and output the index with the greatest running total. Ties choose the lower index.

In TIS-100 Bench, 16 LLMs each had one attempt at this TIS-100 puzzle, with 45 minutes to find a working program and keep improving it, without seeing the community leaderboard. 16 of 16 models solved it. Together they found 7 new Pareto-frontier points beyond the community leaderboard.

16 / 16models solved it
282best LLM cycles (245 community)
7new frontier points

Records

ScoreCommunity bestBest LLMModel
Cycles245282Gemini 4 Argon · +15%
Nodes66DeepSeek V4.1 Flash Ties record
Instructions2837GPT-6 Astra · +32%

Community records from the community leaderboard when we froze it. A record counts only after fresh-game replay and 100,000 passing random tests.

Pareto frontier

Pareto explorer

Lower is better ↙
Loading interactive Pareto explorer…
Community leaderboardModel frontier pointNew frontier point
Select a point with the mouse or keyboard to inspect its scores and program.

Search over time

Loading puzzle trajectories…

Every model

ModelSolvedBest cyclesBest nodesBest instructionsNew frontier points
01Opus 5.5Yes2906373
02Sonnet 5.5Yes3596403
03GPT-6 AstraYes3236372
04Kimi K3.1 ProYes2976391
05GPT-6.1 SolYes3096371
06Gemini 4 ArgonYes282638—
07Gemini 3.8 FlashYes352640—
08Fable 5.1Yes353639—
09Grok 4.7Yes378641—
10GLM-5.3Yes388640—
11DeepSeek V4.1 FlashYes405640—
12Muse Wonder 1.0Yes426640—
13MiMo-V2.6-ProYes456845—
14Muse Spark 1.3Yes561746—
15Haiku 5.5Yes568744—
16GPT-6 LunaYes618846—

Best scores can come from different programs. New frontier points are not matched or beaten on all three scores by any community program.

Best model programs

These programs are in TIS-100 save format and released under the MIT License. All of this puzzle’s model programs are on GitHub and Hugging Face.

Gemini 4 Argon Fewest cycles

282 cycles · 12 nodes · 76 instructionsNEXUS.24.511.7.282-12-76.txt on GitHub

Gemini 4 Argon program running on NEXUS.24.511.7 in TIS-100
Node 0
ADD UP
MOV RIGHT DOWN
MOV ACC DOWN
MOV ACC DOWN
ADD UP
MOV ACC RIGHT
MOV ACC RIGHT
Node 1
ADD UP
MOV ACC LEFT
ADD UP
MOV ACC DOWN
MOV LEFT DOWN
MOV LEFT DOWN
Node 2
ADD UP
MOV RIGHT DOWN
MOV ACC DOWN
MOV ACC DOWN
ADD UP
MOV ACC RIGHT
MOV ACC RIGHT
Node 3
ADD UP
MOV ACC LEFT
ADD UP
MOV ACC DOWN
MOV LEFT DOWN
MOV LEFT DOWN
Node 4
S:MOV UP ACC
SUB UP
JGZ B
MOV UP DOWN
MOV 1 DOWN
JMP S
B:ADD UP
MOV ACC DOWN
MOV 2 DOWN
Node 5
S:MOV UP ACC
SUB UP
JGZ B
MOV UP DOWN
MOV 1 DOWN
JMP S
B:ADD UP
MOV ACC DOWN
MOV 2 DOWN
Node 6
S:MOV UP ACC
SUB UP
JGZ B
MOV UP DOWN
MOV 3 DOWN
JMP S
B:ADD UP
MOV ACC DOWN
MOV 4 DOWN
Node 7
S:MOV UP ACC
SUB UP
JGZ B
MOV UP DOWN
MOV 3 DOWN
JMP S
B:ADD UP
MOV ACC DOWN
MOV 4 DOWN
Node 8
MOV UP RIGHT
Node 9
S:MOV RIGHT ACC
SUB ANY
JGZ R
MOV RIGHT NIL
MOV LAST DOWN
JMP S
R:MOV LAST NIL
MOV RIGHT DOWN
Node 10
MOV UP LEFT
MOV UP LEFT
MOV RIGHT LEFT
MOV RIGHT LEFT
Node 11
MOV UP LEFT

GPT-6.1 Sol Fewest nodes

493 cycles · 6 nodes · 39 instructionsNEXUS.24.511.7.493-6-39.txt on GitHub

GPT-6.1 Sol program running on NEXUS.24.511.7 in TIS-100
Node 0
ADD UP
MOV ACC RIGHT
MOV ACC RIGHT
Node 1
ADD UP
MOV ACC DOWN
MOV LEFT DOWN
MOV LEFT DOWN
MOV RIGHT DOWN
MOV RIGHT DOWN
Node 2
L: ADD UP
SAV
SUB RIGHT
JLZ B
ADD RIGHT
MOV ACC LEFT
MOV 3 LEFT
JMP L
B: MOV RIGHT LEFT
MOV 4 LEFT
SWP
Node 3
ADD UP
MOV ACC LEFT
MOV ACC LEFT
Node 5
L: MOV UP ACC
SUB UP
JGZ T
MOV UP ACC
SUB UP
JLZ R
MOV 1 DOWN
JMP E
T: ADD UP
SUB UP
JLZ R
MOV 2 DOWN
E: MOV UP NIL
JMP L
R: MOV UP DOWN
Node 9
MOV UP DOWN

GPT-6.1 Sol Fewest instructions★ New frontier point

387 cycles · 8 nodes · 37 instructionsNEXUS.24.511.7.387-8-37.txt on GitHub

GPT-6.1 Sol program running on NEXUS.24.511.7 in TIS-100
Node 0
ADD UP
MOV ACC DOWN
MOV ACC DOWN
Node 1
ADD UP
MOV ACC DOWN
Node 2
L: ADD UP
SAV
SUB RIGHT
JLZ B
ADD RIGHT
MOV ACC DOWN
MOV 3 DOWN
JMP L
B: SWP
MOV RIGHT DOWN
MOV 4 DOWN
Node 3
ADD UP
MOV ACC LEFT
MOV ACC LEFT
Node 4
MOV UP RIGHT
Node 5
L: MOV UP ACC
SUB LEFT
JGZ T
MOV LEFT ACC
SUB RIGHT
JLZ R
MOV 1 DOWN
JMP E
R: MOV RIGHT DOWN
JMP L
T: ADD LEFT
SUB RIGHT
JLZ R
MOV 2 DOWN
E: MOV RIGHT NIL
Node 6
MOV UP LEFT
Node 9
MOV UP DOWN