Sapient Open-Sources PRAXIST: 49 Gold Medals on MLE-bench, Beating Claude Opus 4.8
rohanpaul_ai · x · 2026-08-28
Sapient Intelligence released and open-sourced PRAXIST Beta, an autonomous R&D system. It runs parallel "Research Peers" and stores findings, failures, evidence strength, and lineage in a shared research graph, letting later experiments reuse validated mechanisms, retest weak claims, and combine discoveries across branches — carrying experimental evidence across generations.
On the full 75-task MLE-bench, PRAXIST on deepseek-v4-pro earned 49 gold and 60 total medals, beating Claude Opus 4.8's 34 gold / 55 total. In partner environments it achieved a 100% safe-landing rate in rocket simulation and cut accumulated error in an industrial SLAM system from 9.37cm to 5.01cm, with measurable gains in robotics control and quantitative finance.
More from coding & agent
- Vercel Goes Full Backend: Python GA, Containers, and Daemons — cramforce · 2026-10-16
- WorldMemArena Accepted at NeurIPS: Memory Storage ≠ Memory Use — xwang_lk · 2026-10-16
- Vercel Ship Demos Container Deploys for Go, Ruby, Java and More — edwin · 2026-10-16
- TRL adds day-0 support for fine-tuning Muse Spark Open's 816B MoE with LoRA on 64 H100s — QGallouedec · 2026-10-16
- Databricks engineers land impactful PRs on day 2 with AI dev tools, vs week 3 before — Yuchenj_UW · 2026-10-16
- Perplexity open-sources SPACE, the sandbox platform powering its Computer agents — AravSrinivas · 2026-10-16