Coding Agent with Executable World Models Solves 15 of 25 ARC-AGI-3 Benchmark Games
Researchers evaluated a coding-agent system that maintains an executable Python world model to solve ARC-AGI-3 games, achieving full solutions on 15 of 25 public games using GPT-5.5. The system uses verification and refactoring toward simpler abstractions without game-specific hand-coded logic, relying instead on general prompts and a scripted controller. The results suggest that verifier-driven executable world models may be a viable approach for developing more general AI agents capable of abstract reasoning.
A new coding-agent system for the ARC-AGI-3 benchmark maintains an executable Python world model that it verifies against previous observations and refactors toward simpler abstractions as a proxy for minimum description length (MDL) simplicity bias. The agent uses a scripted controller, predefined world-model interfaces, verifier programs, and a plan executor, but contains no hand-coded game-specific logic, heuristics, or hidden solutions—the same agent and prompts work across all games. The researchers conducted a security audit to identify and close unintended information channels that could leak benchmark-specific data. Testing on 25 public ARC-AGI-3 games showed that with GPT-5.5 using high reasoning effort, the agent fully solved 15 games and achieved a mean per-game RHAE (relative human-aligned evaluation) of 58.12%; with GPT-4 high reasoning effort, it solved 8 games with a mean RHAE of 41.29%. The authors note that performance on the private validation set remains untested, and they have released full run artifacts and code for reproducibility.
What's missing
The paper does not discuss performance on the private validation set, which the authors acknowledge is not yet available to them. Additionally, the study does not provide detailed analysis of which game categories or types of reasoning tasks the agent succeeded or failed on, limiting understanding of the approach's strengths and weaknesses.
What different sources said
- arXiv cs.AICenter
Executable World Models for ARC-AGI-3 in the Era of Coding Agents
Related
Gut Bacteria Enzyme Found to Break Down Heat-Processed Food Compounds, Producing Novel Biogenic Amines
Researchers have discovered that an enzyme in common gut bacteria can degrade N-epsilon-carboxymethyllysine (CML), a compound formed during thermal food processing, producing previously unknown biogenic amines. The enzyme, ornithine decarboxylase SpeC from enterobacteria, acts on CML and related modified lysine derivatives through a low-level 'underground' catalytic activity. This finding suggests a previously unrecognized communication axis between thermally processed dietary compounds and gut microbial physiology, with potential implications for host health.
Full-Length Gene Sequencing Reveals Two Distinct Bacterial Communities in Black-Legged Ticks Expanding Into Canada
Researchers used Oxford Nanopore full-length 16S rRNA gene sequencing to characterize the microbiome of Ixodes scapularis black-legged ticks collected in Nova Scotia, Canada, distinguishing between tick-adapted bacteria and environmentally acquired bacteria. The study comes as I. scapularis — the primary vector of Lyme disease — is rapidly expanding northward into Canada due to climate change. The findings suggest that environmentally derived bacteria in tick microbiomes are not mere contamination, which has implications for how tick microbiome data is collected and interpreted across surveillance studies.
Study Identifies Metabolic Link Between Cell Envelope Stress and Biofilm Formation in Bacteria
Researchers have discovered that the metabolite acetyl-CoA directly inhibits enzymes that degrade the bacterial signaling molecule c-di-GMP, connecting cell envelope biosynthesis stress to biofilm formation in Pseudomonas aeruginosa. The study found that sub-inhibitory concentrations of antibiotics targeting early peptidoglycan biosynthesis — but not other antibiotic classes — elevate c-di-GMP levels by reducing phosphodiesterase activity, with acetyl-CoA competing for the enzyme active site. Because the relevant enzyme domain is broadly conserved across bacterial species, this checkpoint mechanism may be widespread and could have implications for understanding antibiotic-induced biofilm responses.