Environments are where agents learn. They define the world, rules, and feedback loop of state → action → reward. Everything from coding/math tasks to games and multi-turn dialogue evals can be thought of as environments. Without them, RL is just math with nothing to interact with.

This is why environments are pivotal to the next wave of AI progress. But while big labs are spending millions buying and privatizing RL environments, open-source has no comparable way to crowd-source them at scale. We’re building the platform and infrastructure to change that.

The Environments Hub is an open, community-powered platform to create, explore, and reuse environments. Lowering the barrier for anyone to contribute to state-of-the-art RL. app.primeintellect.ai/dashbo…

Key features: - Pull, push & manage environments via Hub or CLI - Generate evaluation reports across models - Deep integration with verifiers framework - Native environments support in our prime-rl trainer - Native sandbox support for code execution

In last week’s private beta, over 30 researchers and companies contributed environments to the Hub. Special shoutout to @arcee_ai, @hud_evals, @WhyPhy_Labs, @GroqInc, and the many individuals who contributed their first environments to the hub!

We’re also opening up a list of RFCs and bounties for new environments. These tasks are intentional: collectively building towards INTELLECT-3, a fully open, agentic, state-of-the-art model.

Next Steps: Providing The Full-Stack Open AGI Infrastructure - Crowd-source RL environments at scale - Train a fully open agentic SOTA model with INTELLECT-3 - Seamlessly integrate an ecosystem of tools, sandboxes, RFT, and inference - Provide the open AI infrastructure stack rivaling what’s behind the walls of closed labs

Links - app.primeintellect.ai/dashbo… - docs.primeintellect.ai/tutor… - github.com/PrimeIntellect-ai… - github.com/willccbb/verifier… - verifiers.readthedocs.io/en/… - github.com/primeintellect-ai…