NVIDIA's Skill2Env turns 3.4k Agent Skills into 7,971 RL tasks for $90k+ of GPT-5.6 API
dair_ai · x · 2026-09-24
NVIDIA's Skill2Env compiles public Agent Skills into executable RL training environments.
- Pipeline: a Codex planner reads the SKILL.md bundle, researches public assets, and splits each Skill into workflows; a Codex creator builds each task with programmatic tests and a behavioral rubric drawn from the Skill's own quality criteria
- Scale & cost: from 3.4k crawled Skills, the pipeline produced 7,971 tasks across 13 domains, with software engineering under a quarter of the corpus; generation with GPT-5.6 Sol cost over $90k in API usage
- Results: after 300 steps of outcome-only RL, Qwen3.8-27B improved from 49.4% to 54.1% on Terminal-Bench 2.1 and from 33.4% to 37.7% pass@1 on S2EBench, their hand-verified held-out benchmark
Related event: NVIDIA's Skill2Env Turns 3.4k Agent Skills Into 8k RL Environments(2 posts)→
More from Research
- ArcGrav: a cross-model Agent Control Language to cut token overhead — ArchitecturalGravity · 2026-11-12
- Open-source dual-arm robot solves tangram puzzles with 90 demos, ~60% success — DanGrover · 2026-11-12
- FutureHouse's deslopper trains Kosmos to write indistinguishably from human experts — proteinrosh · 2026-11-12
- Fleuret amplifies debate: AI can't slash wet-lab experiments in unmapped biology — francoisfleuret · 2026-11-12
- Recursive's Eureka Machine ships Flint: a sparse 40B model with 2.5B active for local agentic coding — agihouse_org · 2026-11-12
- Recursive bets on open-endedness: agents running massive parallel experiments like scientist communities — agihouse_org · 2026-11-12