Three ranked leaderboards, each measuring a dimension that raw price alone cannot capture. Scores are computed from the ComputeNav dataset using open methodology — no black box, no sponsored placements. Backed by GCPI Research.
GCPI Efficiency Score
Loading…
Score methodology: min–max of 1 / price_blended over healthy offers with positive price → 0–100.
Data: models.json snapshot —.
GCPI · CC BY 4.0.
EU Compliance Score
Loading…
Score methodology: EU region +50, geo_fencing === EU +30, open weights +20 (max 100). Data:
models.json snapshot —.
GCPI · CC BY 4.0.
Agent Suitability Score
Loading…
Score methodology: latency + throughput + OpenAI-compatible + status (max 100). Excludes
status === "down". Data: models.json snapshot
—. GCPI · CC BY 4.0.