PrimeIntellect-ai / verifiers

Our library for RL environments + evals

camel-ai / loong

🐉 Loong: Synthesize Long CoTs at Scale through Verifiers.

bandr-ai / bandits

Trace mining for post-training. Filter trajectories, draft verifiers, export SFT and RL data.

bluejay-ai-dev / mivas-bench

Multi-Industry Voice Agent Simulation Bench (MIVAS) is an indicator of voice AI performance across economic sectors. MIVAS is comprised of tasks and verifiers with granular rewards executed in production-grade multi-agent RL environments across industries where voice AI is being adopted the fastest.

jinzijian / EvoTrace

Compile real-world Claude Code and Codex trajectories into verified, tradable post-training assets.