AI & data behind Playrift

Playrift mixes your Steam library with public game metadata. Only some flows call a large language model.

Quick Game Pick

When you describe what you want to play, we send your prompt plus algorithm shortlists to OpenRouter using the openai/gpt-4o-mini model. The model must pick from curated candidates so suggestions stay grounded in your taste and RAWG data.

Rift Suggestions & Discovery Radar

The main recommendation grid is not generated by an LLM. Scores come from your Steam playtime and genre signals matched against the RAWG catalog (including multi-platform results). Search uses RAWG's API as well.

Play together & comparisons

Library overlap and co-op picks use the same Steam + RAWG pipeline—rules and catalog data, not chat models.

Models and providers can change as we improve quality and cost; this page reflects the current design of the product.