Public venue rankings
OpenRouter top-50 token rankings and public model list tariffs are snapshotted, versioned, and archived.
Methodology
Observable first. Reproducible by design.
Every print publishes the recipe, the caveats, and the exact scope of its claims, so any reader can recompute the number from the same public inputs.
Tokens priced at list
Observed token volume is valued with a standardized 75/25 input-output blend, then summed by UTC day.
Spend-weighted exposure
Models enter the 14-day basket in spend order until the cumulative spend cutoff is reached.
Tiered claims
Every series names its evidence tier, from public notional (E0) through finance-reconciled (E4).
Eligibility funnel preview
| Stage | Rows | Token share | Note |
|---|---|---|---|
| Raw ranking rows | 51 | n/a | Includes aggregate other row |
| Model rows after excluding other | 50 | 100.00% | OpenRouter top-50 model universe |
| Rows with accepted v1 identifier | 50 | 100.00% | Native permaslug is the v1 join key |
| Rows with public tariff | 49 | 99.63% | 1 model(s) lack a tariff |
| Coverage-qualified print | 49 | 99.63% | Above 50% validation threshold |
| 95% trailing-spend basket | 17 | 53.58% | Captures 95.14% of trailing spend |
Scope note
Public notional, not realized spend.
The series values public token volumes at public list tariffs. Discounts, caching, private terms, and non-OpenRouter traffic sit outside the headline print until higher evidence tiers are added.
Diagnostics
Input integrity
Forensic screens on the source panel.
The pooled model-day token panel runs through Benford digit tests, last-two-digit uniformity, and exact-duplication analysis. Current results: close conformity on all digit tests, zero duplicates, and zero restatements in the archived history. These are screens for fabricated or smoothed source data; provenance attestation sits at the higher evidence tiers.
Error model
The gap to realized spend.
The difference between realized and notional spend decomposes into five documented terms: the 75/25 blend convention (bounded by the published band), provider routing below list, discounts and caching, the top-50 coverage cutoff, and model-to-tariff mapping error.
Governance
Who publishes this, and under what rules.
The Grid publishes the numbers, discloses its conflict of interest as a market operator, and makes every print reproducible from public inputs. The governance page carries the full benchmark statement and policies.
Public notional
Public volumes x public list tariffs
LiveHistorical public notional
Public volumes x dated historical tariffs
LiveTransaction-priced venue
Consumed venue units at executed prices
RoadmapAttested venue aggregate
Venue aggregates with source controls
RoadmapFinance-reconciled realized
Spend reconciled to invoices or ledgers
RoadmapNo silent edits
A corrected print is a new version with a new hash; prior versions remain archived and material corrections are announced.
Announced effective dates
Methodology changes take effect prospectively. Pro forma backcasts are published as labeled research series.
No double counting
Multi-source expansion comes with announced eligibility rules, non-overlap checks, transition reporting, and independent review of self data.
Names that outlive contracts
Renamed series map old identifiers to new in the manifest with no numeric break. Discontinuation carries a minimum 90-day notice, a stated fallback where one exists, and permanent archival of all published prints.
Complaints and corrections
Complaints, correction reports, and citation questions go to [email protected].
Data access
Request data Build against the print.
Download the public artifacts, request the recomputation package, or talk to The Grid about moving from public notional to verified spend.