Post by Scale AI

We’re heading to ICLR 2026 with six accepted papers. 🎉 This research spans agentic RL, LLM evaluation, and AI safety — all core to building reliable AI systems. Take an early look at what we’ll be presenting later this month ⬇️