TIS-100 Bench: Signal Error Corrector
Copy two streams, replacing a negative value with its partner from the other stream. At most one partner is negative at a time.
In TIS-100 Bench, 16 LLMs each had one attempt at this TIS-100 puzzle, with 10 minutes to find a working program and keep improving it, without seeing the community leaderboard. 16 of 16 models solved it. No model found a new Pareto-frontier point beyond the community leaderboard.
Records
| Score | Community best | Best LLM | Model |
|---|---|---|---|
| Cycles | 138 | 201 | Gemini 4 Argon · +46% |
| Nodes | 5 | 5 | DeepSeek V4.1 Flash Ties record |
| Instructions | 13 | 16 | Gemini 4 Argon · +23% |
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 ↙Search over time
Loading puzzle trajectories…
The search over time
Best in-game score so far, before independent testing. Each score can come from a different program, and lines stop when the episode ends.
Every model
| Model | Solved | Best cycles | Best nodes | Best instructions | New frontier points |
|---|---|---|---|---|---|
| 01Gemini 4 Argon | Yes | 201 | 5 | 16 | — |
| 02Muse Wonder 1.0 | Yes | 233 | 5 | 19 | — |
| 03Gemini 3.8 Flash | Yes | 236 | 5 | 17 | — |
| 04Sonnet 5.5 | Yes | 245 | 6 | 20 | — |
| 05Fable 5.1 | Yes | 278 | 5 | 17 | — |
| 06GPT-6 Astra | Yes | 278 | 5 | 18 | — |
| 07GPT-6.1 Sol | Yes | 278 | 5 | 17 | — |
| 08Opus 5.5 | Yes | 278 | 5 | 17 | — |
| 09Haiku 5.5 | Yes | 307 | 6 | 20 | — |
| 10Kimi K3.1 Pro | Yes | 329 | 5 | 18 | — |
| 11Grok 4.7 | Yes | 350 | 6 | 20 | — |
| 12Muse Spark 1.3 | Yes | 350 | 5 | 20 | — |
| 13DeepSeek V4.1 Flash | Yes | 357 | 5 | 18 | — |
| 14GLM-5.3 | Yes | 378 | 6 | 24 | — |
| 15GPT-6 Luna | Yes | 388 | 5 | 19 | — |
| 16MiMo-V2.6-Pro | Yes | 398 | 5 | 20 | — |
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
201 cycles · 6 nodes · 29 instructionsNEXUS.03.176.9.201-6-29.txt on GitHub

MOV UP, ACC
MOV ACC, RIGHT
MOV ACC, DOWNS: MOV UP, ACC
JGZ O
MOV LEFT, DOWN
MOV UP, ACC
JGZ O
MOV LEFT, DOWN
MOV UP, ACC
JGZ O
MOV LEFT, DOWN
JMP S
O: MOV LEFT, NIL
MOV ACC, DOWNS: MOV UP, ACC
JGZ O
MOV RIGHT, DOWN
MOV UP, ACC
JGZ O
MOV RIGHT, DOWN
JMP S
O: MOV RIGHT, NIL
MOV ACC, DOWNMOV UP, ACC
MOV ACC, LEFT
MOV ACC, DOWNMOV UP, DOWNMOV UP, DOWNGemini 4 Argon
274 cycles · 5 nodes · 17 instructionsNEXUS.03.176.9.274-5-17.txt on GitHub

MOV UP, ACC
JGZ O
MOV RIGHT, ACC
MOV ACC, DOWN
JMP P
O: MOV RIGHT, DOWN
P: MOV ACC, DOWNMOV UP, LEFTMOV UP, DOWNMOV UP, ACC
JGZ O
MOV UP, ACC
MOV ACC, DOWN
JMP P
O: MOV UP, DOWN
P: MOV ACC, RIGHTMOV LEFT, DOWNGemini 4 Argon
360 cycles · 5 nodes · 16 instructionsNEXUS.03.176.9.360-5-16.txt on GitHub

MOV UP, ACC
JGZ A
MOV RIGHT, ACC
MOV ACC, DOWN
JMP E
A: SAV
MOV ACC, DOWN
MOV RIGHT, ACC
JGZ E
SWP
E: MOV ACC, DOWNMOV UP, LEFTMOV UP, DOWNMOV UP, DOWN
MOV UP, RIGHTMOV LEFT, DOWN