New Method Scores on ARC-AGI-3
bryanlanders · x · 2026-07-17
This repost highlights that follow-up research around ARC-AGI-3 is accelerating. Just 3.5 months after V3's release, the ecosystem is already showing a snowball effect.
The cited content introduces a harness named schema: on the ARC-AGI-3 Public set, combining Opus 4.8 + Fable 5 achieves 99% RHAE, while combining GPT-5.6 Sol reaches 95.35%. The author claims this system enables LLMs to "think like physicists," making it highly worthwhile to break down its methodology and reproduction details in the future.
Related event: Schema Harness Sparks ARC-AGI-3 Debate(14 posts)→
More from Research
- Hierarchical continuous flow language models make pathwise RL more robust to critic exploitation — vxnuaj · 2026-10-27
- rauchg reproduces OpenAI's covert CPU-channel finding: chat across Firecracker microVMs with no network — ccerrato147 · 2026-10-27
- Microsoft Research: ForkPilot cuts long-horizon agent compute by up to 59% — EliasEskin · 2026-10-27
- Microsoft's TeleTune evolves agent skills from offline user telemetry — EliasEskin · 2026-10-27
- FlexGlue: sparse-to-dense matching that beats dense accuracy at a fraction of the cost — zhenjun_zhao · 2026-10-27
- PixGR: pixel-space generative decoding for fast feed-forward 3D Gaussian reconstruction — zhenjun_zhao · 2026-10-27