AIHacker Newsabout 2 hours ago
16 LLMs Solve Zachtronics TIS-100 Game
11 min read
Researchers tested sixteen large language models on the programming game TIS-100, revealing that Opus 5.5 and GPT-6 Astra solved all 55 puzzles. Opus 5.5 found 72 new frontier points versus 32 for GPT-6 Astra. The study used a 10 to 45 minute time limit per puzzle and required 100,000 passing tests to verify generalization.
Level
Hype check
Tap to vote and see what everyone thinks.