SpaceChem Bench: Of Pancakes and Spaceships
research_example_1 Easy Bonding & routing Single-reactor Sernimir II · 
SpaceChem Bench report →
Inputs: α Oxygen (O2). Output: ψ Oxygen (O2) ×10.
In SpaceChem Bench, 16 LLMs each had one attempt at this SpaceChem puzzle, with 10 minutes to build a working solution and keep optimizing it, without seeing the community archive. 16 of 16 models solved it. No model found a new Pareto-frontier point beyond the frozen community archive.
Records
| Score | Community best | Best LLM | Model |
|---|---|---|---|
| Cycles | 45 | 45 | GPT-6 Astra Ties record |
| Reactors | 1 | 1 | DeepSeek V4.1 Flash Ties record |
| Symbols | 6 | 6 | DeepSeek V4.1 Flash Ties record |
Community records from the community archive when we froze it on October 18, 2026. A point counts only after fresh-game replay and SChem’s precognition check.
Pareto frontier
Pareto explorer
Lower is better ↙Search over time
Loading puzzle trajectories…
The search over time
Best in-game score so far, before replay. Each score can come from a different solution, and lines stop when the episode ends.
Every model
| Model | Solved | Best cycles | Best reactors | Best symbols | New frontier points |
|---|---|---|---|---|---|
| 01GPT-6 Astra | Yes | 45 | 1 | 6 | — |
| 02GPT-6.1 Sol | Yes | 45 | 1 | 6 | — |
| 03Gemini 4 Argon | Yes | 52 | 1 | 6 | — |
| 04GLM-5.3 | Yes | 53 | 1 | 10 | — |
| 05Opus 5.5 | Yes | 55 | 1 | 6 | — |
| 06Gemini 3.8 Flash | Yes | 57 | 1 | 6 | — |
| 07Kimi K3.1 Pro | Yes | 58 | 1 | 6 | — |
| 08Muse Spark 1.3 | Yes | 58 | 1 | 6 | — |
| 09Muse Wonder 1.0 | Yes | 58 | 1 | 6 | — |
| 10Grok 4.7 | Yes | 59 | 1 | 6 | — |
| 11Fable 5.1 | Yes | 61 | 1 | 6 | — |
| 12Sonnet 5.5 | Yes | 63 | 1 | 6 | — |
| 13DeepSeek V4.1 Flash | Yes | 65 | 1 | 6 | — |
| 14GPT-6 Luna | Yes | 115 | 1 | 6 | — |
| 15MiMo-V2.6-Pro | Yes | 115 | 1 | 8 | — |
| 16Haiku 5.5 | Yes | 116 | 1 | 6 | — |
Best scores can come from different solutions. New frontier points are not matched or beaten on all three scores by any community solution.
Best model solutions
These solutions are in SpaceChem Community Edition export format and released under the MIT License. All of this puzzle’s model solutions are on GitHub and Hugging Face.
GPT-6 Astra
45 cycles · 1 reactor · 14 symbolsresearch_example_1.45-1-14.txt on GitHub
Layout
Running in the game
SpaceChem Community Edition export (research_example_1.45-1-14.txt)
SOLUTION:Of Pancakes and Spaceships,Player,0-0-0
COMPONENT:'empty-research-reactor',2,0,''
MEMBER:'instr-start',-90,0,128,2,5,0,0
MEMBER:'instr-start',-90,0,32,2,2,0,0
MEMBER:'instr-arrow',0,0,64,2,1,0,0
MEMBER:'instr-arrow',180,0,64,6,1,0,0
MEMBER:'instr-grab',-1,0,128,2,1,0,0
MEMBER:'instr-input',-1,0,128,2,4,0,0
MEMBER:'instr-rotate',-1,0,128,3,1,0,0
MEMBER:'instr-output',-1,0,128,5,1,0,0
MEMBER:'instr-grab',-1,0,128,6,1,0,0
MEMBER:'instr-arrow',0,0,16,2,1,0,0
MEMBER:'instr-arrow',180,0,16,6,1,0,0
MEMBER:'instr-grab',-1,0,32,2,1,0,0
MEMBER:'instr-output',-1,0,32,3,1,0,0
MEMBER:'instr-input',-1,0,32,4,1,0,0
MEMBER:'instr-rotate',-1,0,32,5,1,0,0
MEMBER:'instr-grab',-1,0,32,6,1,0,0
DeepSeek V4.1 Flash
115 cycles · 1 reactor · 6 symbolsresearch_example_1.115-1-6.txt on GitHub
Layout
Running in the game
SpaceChem Community Edition export (research_example_1.115-1-6.txt)
SOLUTION:Of Pancakes and Spaceships,Player,0-0-0
COMPONENT:'empty-research-reactor',2,0,''
MEMBER:'instr-start',0,0,128,0,1,0,0
MEMBER:'instr-arrow',0,0,64,1,1,0,0
MEMBER:'instr-input',0,0,128,1,1,0,0
MEMBER:'instr-grab',0,0,128,2,1,0,0
MEMBER:'instr-grab',0,0,128,7,1,0,0
MEMBER:'instr-arrow',180,0,64,7,1,0,0
MEMBER:'instr-start',180,0,32,1,0,0,0
MEMBER:'instr-output',0,0,32,0,0,0,0