Terminal-Bench-RL
Training long-horizon terminal agents with RL. GRPO training code which scales to 32xH100s for long horizon terminal/coding tasks. Base agent is now the top Qwen3 agent on Stanford's TerminalBench leaderboard. After training calculator agent via RL, I really wanted to go bigger!
Open source
Reactions (1)
Similar products
by description similarity
ART
github.coma new open-source RL framework for training agents
Superset
superset.shTerminal to run 10 parallel coding agents
NanoRL
github.comRL training for LLMs in ~1,800 lines
RULER
openpipe.aiEasily apply RL to any agent
Coding Agent singularly focused efficiency and context curation. Reduces API costs by 50-80% vs other agent AND improves the code quality at the same time. Uses Hash Anchored edits, massively parallel operations, AST manipulation and many many other optimizations. https://dirac.run/