Team hits 70% on ARC-AGI-2 Kaggle with 4 L4 GPUs and no internet
JFPuget · x · 2026-08-19
JFPuget and teammate (team NVBANANA) passed 70% in the ARC-AGI-2 competition on Kaggle, running without internet access on just 4x L4 GPUs. He notes the code would run much faster on an H100, lowering cost per puzzle. Kaggle uses the same test data as the official leaderboard, so the score is directly comparable to frontier models.
Related event: Team NVBANANA Tops 70% on ARC-AGI-2 Offline with Just 4 L4 GPUs(2 posts)→
More from Models
- Grok Bot Ran a 12-Hour Data Analysis Nonstop, Handling API Rate Limits Itself — Daniel_Farinax · 2026-10-16
- One year of Recursive Language Models: GPT-5-mini beats GPT-5 by 110% on 132k-token tasks, cheaper too — a1zhang · 2026-10-16
- Users Report Claude Models Suddenly More Compliant After Forced Compaction — RealSharpNinja · 2026-10-16
- Prediction: Opus 5.5-level models on laptops by 2027, backed by Qwen 27B data — cephaloform · 2026-10-16
- Users Say Krea 2 Output Degraded Sharply Since Krea 3 Beta Launched — szywis · 2026-10-16
- LLM-as-a-Verifier lifts Nemotron 3.5 Super by 9-10 points on agentic benchmarks — simonguozirui · 2026-10-16