prime-rl
Prime Intellect’s Apache-2.0 asynchronous RL training framework, installed from git/uv (no verified PyPI package) and integrated with verifiers environments.
Tags
Description
Established Prime Intellect project (repo PrimeIntellect-ai/prime-rl; not archived; not presented as a brand-new 2026 launch). Root LICENSE is Apache-2.0. PyPI packages named prime-rl and prime_rl returned HTTP 404 on 2026-11-12 and again on 2026-11-13; this entry does not invent an installable PyPI name. GitHub latest release v0.10.0 was published 2026-10-13T22:20:01Z (not draft or prerelease). README documents a manual install (git clone, submodules, uv sync) and a one-line curl of scripts/install.sh (that script was not executed). It also states you currently need at least one NVIDIA GPU, and it points to Prime Intellect’s compute platform for renting GPUs — that is a paid vendor service and was not used. In-repo docs live under docs/ (overview, training, eval, scaling, and others). Native integration with verifiers and the Environments Hub is claimed in the README. Training stacks can execute model and environment code at scale. Same operator family as verifiers.
Agent access
Related
- verifiers: Prime Intellect’s MIT library for creating LLM reinforcement-learning environments and evals, with a matching PyPI package.
- OpenEnv: BSD-3-Clause framework for building and serving isolated agentic RL environments over a Gymnasium-style reset/step/state API.
- verl: Distributed reinforcement-learning training library for LLMs, with GRPO/PPO dataflows and pluggable training and rollout engines for agentic post-training.
- SkyRL: Modular reinforcement-learning stack with a training engine, long-horizon agent layer and Gymnasium-style tool-use environments for LLMs.
- TRL: Hugging Face Transformer Reinforcement Learning library for post-training methods such as SFT, DPO, PPO and GRPO on top of transformers.
Guides & comparisons
- RL environment libraries compared — NeMo Gym, OpenEnv and Verifiers (Comparison): Compare NeMo Gym, OpenEnv and Verifiers as libraries for writing and running RL environments: authoring model, isolation, documented trainers, install paths, licences and current versions. Distinct from Harbor/AgentEnv/Seahaven and from the agent RL stacks comparison.
Sources
- https://github.com/PrimeIntellect-ai/prime-rl
- https://github.com/PrimeIntellect-ai/prime-rl/blob/main/LICENSE
- https://github.com/PrimeIntellect-ai/prime-rl/blob/main/README.md
- https://github.com/PrimeIntellect-ai/prime-rl/blob/main/docs/overview.md
- https://github.com/PrimeIntellect-ai/prime-rl/releases/tag/v0.10.0
Machine-readable
- JSON
- Markdown
- Source file on GitHub (edit via pull request)