SERVICE ASSESSMENTS

API service ratings

Overall ratings range from 0 to 5 stars; dimensions range from 0 to 100. Stars are shown only for decisions that use calibrated public rules, have all five required samples and meet their scope and validity requirements. Missing values are not zero, and weights are not redistributed.

Stability

Successful real requests for the same model, access provider, region and workload. Website checks and official status claims are recorded separately.

Response time

Request latency with documented test conditions and sample windows. Website HEAD request duration is not time to first token.

Pricing and fees

Quotes or bills matched by currency, unit, input/output mix, processing tier and applicable conditions.

Models and features

Completed protocol and capability tests with valid samples. A listed model name does not establish that a feature works.

Service guarantees

Real support cases assessed against the service’s commitments. Contact details, business type and marketing copy do not earn points by themselves.

Each valid rating includes weights, sample counts and windows, raw metrics, model/provider/region/workload, costs, validity, rule version and calibration evidence. Different scopes are not ranked together. Promotions do not change scores; serious risks are flagged separately and suspend recommendations and promotions.