TIS-100 Bench: Sequence Merger

Merge two already-sorted, zero-terminated sequences. Advancing one input must preserve the unconsumed value from the other.

In TIS-100 Bench, 16 LLMs each had one attempt at this TIS-100 puzzle, with 30 minutes to find a working program and keep improving it, without seeing the community leaderboard. 12 of 16 models solved it. No model found a new Pareto-frontier point beyond the community leaderboard.

12 / 16models solved it
347best LLM cycles (279 community)
0new frontier points

Records

ScoreCommunity bestBest LLMModel
Cycles279347GPT-6 Astra · +24%
Nodes66Fable 5.1 Ties record
Instructions2225GPT-6.1 Sol · +14%

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
01GPT-6 AstraYes347627—
02Opus 5.5Yes361629—
03Sonnet 5.5Yes361632—
04Fable 5.1Yes371629—
05GPT-6.1 SolYes372625—
06Haiku 5.5Yes398637—
07Gemini 3.8 FlashYes399630—
08Muse Wonder 1.0Yes445640—
09Kimi K3.1 ProYes456633—
10Gemini 4 ArgonYes465629—
11Muse Spark 1.3Yes620648—
12GPT-6 LunaYes656641—
13DeepSeek V4.1 FlashNo————
14GLM-5.3No————
15Grok 4.7No————
16MiMo-V2.6-ProNo————

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.

GPT-6 Astra Fewest cycles

347 cycles · 7 nodes · 34 instructionsNEXUS.00.526.6.347-7-34.txt on GitHub

GPT-6 Astra program running on NEXUS.00.526.6 in TIS-100
Node 1
MOV UP DOWN
Node 2
MOV UP DOWN
Node 3
R: MOV UP ACC
JEZ Z
L: SAV
SUB RIGHT
JLZ B
MOV 9 RIGHT
SWP
JMP L
Z: MOV RIGHT NIL
MOV 2 RIGHT
JMP R
B: SWP
MOV 1 RIGHT
MOV ACC DOWN
Node 4
R: MOV RIGHT ACC
L: JNZ P
MOV 999 LEFT
JMP W
P: MOV ACC LEFT
W: JRO LEFT
JMP L
D: MOV ACC DOWN
JEZ R
MOV RIGHT ACC
MOV ACC DOWN
JEZ R
MOV RIGHT ACC
JMP D
MOV ACC DOWN
Node 5
MOV UP LEFT
Node 7
MOV UP RIGHT
Node 8
MOV ANY DOWN

Opus 5.5 Fewest nodes

361 cycles · 6 nodes · 31 instructionsNEXUS.00.526.6.361-6-31.txt on GitHub

Opus 5.5 program running on NEXUS.00.526.6 in TIS-100
Node 1
MOV UP, ACC
JNZ K
MOV 997, ACC
K: SUB 995
MOV ACC, DOWN
Node 2
MOV UP, ACC
JNZ K
MOV 998, ACC
K: SUB 995
MOV ACC, DOWN
Node 3
MOV UP, RIGHT
Node 4
T: MOV LEFT, ACC
U: SAV
SUB RIGHT
JLZ A
MOV 1, RIGHT
SWP
JMP U
A: SWP
MOV ACC, RIGHT
MOV ACC, DOWN
Node 5
K: JEZ S
L: MOV ACC, LEFT
JRO LEFT
MOV ACC, DOWN
S: MOV UP, ACC
Node 8
MOV ANY, ACC
SUB 2
JEZ Z
ADD 997
Z: MOV ACC, DOWN

GPT-6.1 Sol Fewest instructions

457 cycles · 6 nodes · 25 instructionsNEXUS.00.526.6.457-6-25.txt on GitHub

GPT-6.1 Sol program running on NEXUS.00.526.6 in TIS-100
Node 1
MOV UP DOWN
Node 2
MOV UP DOWN
Node 3
MOV UP RIGHT
Node 4
R: MOV LEFT ACC
JMP C
B: MOV 1 RIGHT
SWP
C: SAV
JNZ S
MOV 999 ACC
S: SUB RIGHT
JGZ B
SWP
MOV ACC DOWN
ADD 2
MOV ACC RIGHT
Node 5
R: MOV UP ACC
JNZ P
MOV 999 ACC
P: MOV ACC LEFT
JRO LEFT
MOV ACC DOWN
JMP R
JMP P
Node 8
MOV ANY DOWN