Prime Intellect Releases IPO Algorithm as Default for Asynchronous RL

Prime Intellect released prime-rl v0.10.0, introducing IPO (INTELLECT Policy Optimization) as the default algorithm for asynchronous RL, using a new token masking scheme to fix probability mismatches and crashes previously seen with IcePop.

2026-10-20 ~ 2026-10-21 · 4 related posts

• Prime Intellect ships prime-rl v0.10.0 with async-stable IPO as default RL algorithm — xeophon · 2026-10-20
• Prime Intellect unveils INTELLECT Policy Optimization, fixing IcePop's ratio masking — xeophon · 2026-10-20
• Prime Intellect's IPO masking fixes IcePop collapse in agentic SWE RL on GLM-4.5-Air — nrehiew_ · 2026-10-20

1 near-duplicate retellings: misovalko