# Clanker Cloud Evals

Clanker Cloud publishes benchmark framing for frontier labs and mid-sized AI research teams evaluating model and agent performance on realistic infrastructure tasks.

## Current focus

- Real infrastructure tasks instead of toy prompts.
- GPU-oriented workflows on preferred providers.
- Model and agent comparisons that reflect operational constraints.
- Evaluation framing for deployment, training, and runtime workflows.

## Notes

- The public evals page is the main surface for charts, tables, and methodology context.
- Dataset-style markup is published on the HTML page for the evals surface.

## Related page

- [Open the evals page](https://clankercloud.ai/evals)