Anthropic, Cohere, DeepSeek, Google Gemini, Groq, OpenAI, OpenRouter, xAI
Transparency
AI Pricing Hub Trust Center
AI Pricing Hub Trust Center explains coverage, update pipeline, data quality, methodology, API status, limitations, and validation for the static AI pricing dataset.
Coverage
What AI Pricing Hub monitors
Validated current models available for cost surfaces.
534 tracked model histories.
United States, Spain, Brazil, and Portugal via language-market configuration.
English, Spanish, and Portuguese public page variants.
OpenAPI spec version 1.0.0.
Generated pricing dataset timestamp.
Derived from the committed pricing dataset used for this static build.
Sitemap is generated after Trust Center during the same static build.
Update pipeline
Daily workflow from provider data to search engines
- 1Provider
Provider source pages and static seed data are refreshed by scheduled update scripts.
- 2Normalization
Raw provider names, model ids, context windows, and price fields are normalized into providers.json.
- 3Validation
Pricing quality checks exclude incomplete cost rows from rankings, calculators, comparisons, and workload recommendations.
- 4Historical snapshots
Daily provider history files preserve first seen, latest seen, and price changes by model.
- 5SEO generation
Static model, provider, comparison, history, ranking, workload, API, and market pages are generated.
- 6Deployment
Cloudflare Pages serves the static build after GitHub workflow validation on the configured branch.
- 7Search engines
Sitemap and robots rules expose indexable HTML pages and exclude raw API/data assets.
Data quality
Validation rules and latest status
Excluded from cost calculations, rankings, comparisons, and workload recommendations.
Rows excluded by shared pricing validation or incomplete metadata policy.
Pricing quality report timestamp.
487 of 487 models expose complete required price fields.
SEO and crawl-budget validators fail the build on missing internal targets.
0 critical pricing errors.
| Rule | Policy |
|---|---|
| Required pricing | Input and output prices must be finite non-negative numbers for cost surfaces. |
| Zero pricing | Zero prices are rejected unless explicitly verified as free. |
| Preview/inactive rows | Incomplete preview or inactive rows are reported and excluded from rankings. |
| Source freshness | Rows retain source status and source URL metadata where available. |
| Build failure policy | No critical pricing issues in the latest validation run. |
Historical coverage
History charts from provider snapshots
Methodology
How AI Pricing Hub works
Provider pricing is collected into generated static JSON, normalized to model/provider rows, and checked before any cost page is generated.
Rankings only use rows that pass shared pricing validation and are sorted by deterministic pricing, stability, recency, or trend rules.
Comparison pages pair meaningful models and normalize cost examples with the same request and token assumptions.
The calculator multiplies input, cached input, output, request volume, days, and optional batch pricing using the same validated pricing model.
Workload pages filter validated models by visible metadata such as context, modalities, provider, and pricing fit.
API status
Public developer platform status
Current public REST API version.
OpenAPI paths exposed under /api/v1.
Generated from static datasets, no runtime database.
Developer docs and Swagger UI are generated statically.
Transparency
Known limitations and missing data policy
- AI Pricing Hub tracks pricing and metadata, not quality benchmarks or latency.
- Provider pages can change without notice between daily refreshes.
- Capability labels are conservative when a field is not explicit in the available data.
62 preview or experimental models are detected from current metadata. Incomplete preview pricing is reported and excluded from cost surfaces.
Missing fields are shown as not listed. AI Pricing Hub does not convert missing prices or zero values into free pricing unless a verified-free signal is present.
FAQ
Trust Center FAQ
How often is AI Pricing Hub updated?
The data pipeline is designed for daily provider updates, validation, history snapshots, static page generation, and deployment.
What data is included in the Trust Center?
The Trust Center uses the generated pricing dataset, history snapshots, OpenAPI specification, sitemap, and validation reports created during the static build.
Why are some models excluded from rankings?
49 models are excluded when they are incomplete, inactive, preview-only with missing required pricing, or fail shared pricing validation.
How many models are currently active?
The current build has 438 active priced models across 8 monitored providers.
Does the API require authentication?
No. Public API v1 is unauthenticated and exposes generated static datasets through JSON endpoints.
How should missing pricing be interpreted?
Missing pricing means the source-backed metadata does not include the required value. AI Pricing Hub does not infer free pricing from missing or zero fields.
Continue
Pick up where you left off
Public API
Build with AI Pricing Hub data
Use static JSON endpoints for providers, models, rankings, history, market metrics, and changelog events.
Editorial information
Reviewed by AI Pricing Hub Editorial
2026-08-04
Methodology explains collection, validation, limitations, and update cadence.