Prime Intellect

The GyReLL environment is published on the Environments Hub as lumpenspace/gyrell.

Use it

# with the prime CLI (https://docs.primeintellect.ai/cli)
prime env install lumpenspace/gyrell

# evaluate any OpenAI-compatible model on it
uv run vf-eval gyrell -m openai/gpt-4.1-mini -n 5

Train on it

Ready-made prime-rl configs live in configs/ — rl-smoke (sanity run), rl-selfplay, eval-checkpoint, and dryrun-free (free-tier models end to end):

prime train configs/rl-smoke.toml -e PRIME_API_KEY -y

Publish changes

The environment package is environments/gyrell/; it vendors a snapshot of src/turngames so the wheel is self-contained.

environments/gyrell/sync_engine.sh    # refresh the vendored engine
cd environments/gyrell
prime env push                        # bump version in pyproject.toml first

Old hub id

The env previously shipped as lumpenspace/codewords; that id stays published so existing runs keep resolving. New work targets lumpenspace/gyrell.