ModelsAug 28, 2026 · 15:423 sources
Sapient's PRAXIST Beta Beats Claude Opus 4.8 on MLE-Bench in Open Source Launch
From huggingnews.com
Sapient Intelligence released PRAXIST Beta, an open-source autonomous R&D system that uses parallel "Research Peers" to explore competing approaches and share findings via a research graph. The system achieved 49 gold medals and 60 total medals across 75 MLE-bench tasks using deepseek-v4-pro, surpassing the 34 gold and 55 total medals reached by Claude Opus 4.8, and operates at roughly 1/12th the token cost of a Claude Code and Opus 4.8 baseline. In partner environments, PRAXIST reached a 100% safe-landing rate in a rocket simulation and reduced accumulated error in an industrial SLAM system from 9.37 centimeters to 5.01 centimeters.
Read the sourceSign in to suggest
Sources
#open-source#agent#research#codingMarkdown
Content history (2)
- 2026-10-16 · Summary · vi · Replaced stored textSapient Intelligence vừa phát hành PRAXIST Beta, một hệ thống R&D tự hành mã nguồn mở. Hệ…Sapient Intelligence đã ra mắt PRAXIST Beta, một hệ thống nghiên cứu và phát triển tự độn…
- 2026-10-16 · Title · vi · Replaced stored textHệ thống mã nguồn mở PRAXIST Beta của Sapient vượt mặt Claude Opus 4.8 trên MLE-BenchPRAXIST Beta của Sapient vượt trội Claude Opus 4.8 trên MLE-Bench trong ra mắt mã nguồn mở
One ranked edition, every morning
The stories that matter today. Free, cancel anytime.
Telegram @aidr_today