OpenRSI-Index v0.1 launches to benchmark recursive AI self-improvement on 1k-GPU clusters
ChengleiSi · x · 2026-09-24
- OpenRSI and the research community released OpenRSI-Index v0.1, an open standard measuring whether AI research agents can recursively improve beyond human-designed training recipes on production-scale clusters.
- Approach: fully open-source projects become autoresearch environments, with agent trajectories lasting 60+ hours; building v0.1 took 100K+ H100-hours.
- Tasks cover real model development: pre-training, post-training, and vision generation; featured runs used up to 20,864 H100-hours each.
- Goal: test whether agents can discover better methods and genuinely extend the scientific and intelligence frontier.
More from Research
- Researchers find reasoning models produce fractals when solving hard problems like ARC-AGI — ShahabBakht · 2026-09-24
- Optimization hardness as transient chaos: revisiting a classic Nature Physics k-SAT paper — wgilpin0 · 2026-09-24
- Transluce releases 30,000 logs tracing rogue AI agent hacking back to March — JacobSteinhardt · 2026-09-24
- Ex-TikTok MLE: JEV could revolutionize recommendation and search systems — hardimanjames · 2026-09-24
- WeightWatcher K-matrix power-law exponent tracks LLM memorization with ρ ≈ -0.90 — rickasaurus · 2026-09-24
- Researchers launch Prevent, Contain, Prove, a voluntary framework for formal methods — Miles_Brundage · 2026-09-24