QUICKSTART¶

Getting the software running locally.

Installation¶

RUNE is a Python-based platform. The core package can be installed with optional extras.

pip install rune-bench                  # core only
pip install rune-bench[holmes]          # + holmesgpt + supabase
pip install rune-bench[all]             # + all optional extras (holmesgpt, vastai-sdk)

Running Your First Benchmark¶

The simplest way to run a benchmark is using existing Ollama server mode:

# Set your existing Ollama server URL

export RUNE_OLLAMA_URL=http://localhost:11434


# Run a benchmark with a specific model and question

python -m rune run-benchmark \
  --model llama3.1:8b \
  --question "Why is the cluster unhealthy?"

Running with Vast.ai Provisioning¶

To provision a GPU instance on Vast.ai:

export VAST_API_KEY=...

python -m rune run-benchmark \
  --vastai \
  --question "What is the status of the deployment?"

Docker Quickstart¶

# Build image

docker build -t ai-benchmark-rune .


# Run existing server mode

docker run -it --rm \
  ai-benchmark-rune run-ollama-instance \
  --ollama-url http://host.docker.internal:11434

Local Development Setup¶

For full environment setup, build/test/lint commands per repo, and Definition of Done validation steps, see the Developer Guide.