Astra reportedly saturates ARC-AGI-3 with a 99.9% score
smlpth · x · 2026-09-04
A post cites claims that the agent Astra has saturated ARC-AGI-3 with a 99.9% score. The author recalls a Dwarkesh Patel–François Chollet conversation from two years ago, when he too doubted LLMs could ever crack ARC-AGI 1 — a reminder of how quickly the benchmarks keep falling. Figure is unverified secondhand.
More from Models
- Opus 5.5 tops OpenRouter spending in 20 of 29 task categories — mmurph · 2026-10-21
- Meta team builds open-source 30B LLM from scratch in months, hails 'second era' — TimDarcet · 2026-10-21
- DFlash drafter for Muse Glimmer-30B: 5 layers, blockwise 16-token spec decoding — TimDarcet · 2026-10-21
- EvoPress quantization shrinks 30B model from 56GB to 19GB at near-bf16 accuracy — TimDarcet · 2026-10-21
- Radar chart: Muse Glimmer-30B benchmarked against Gemma4-31B and Qwen3.6-27B — TimDarcet · 2026-10-21
- Privacy evals: Muse Glimmer-30B leaks less than Qwen3.6-27B on CIMemories — TimDarcet · 2026-10-21