The Token Volume Index

Settled daily token volume across the 11 OTPI labs: share within a weight class, a shared growth baseline, an open/closed split, and the closed-lab REST series.

The Token Volume Index counts the tokens OTPI already prices. Nothing here is a new collection — the series are derived from the same settled daily volumes. Open-weight models released by a closed lab (gpt-oss, Gemma) and free variants are excluded before settlement, so a lab’s totals sit below its published all-model figures.

Daily grain only. No Index window: every request needs a Premium or Full key. See Access tiers.

Five charts on the platform

The page at data.ornn.com/analytics/models/volume draws five charts: four from the settled volume feed, and Input vs Output Share from the token-split feed. Token Share and Volume Growth each have an independent selector that gates on weight class first — closed source or open source — then the labs inside that class, and both open on the closed frontier three (anthropic, openai, google). The two classes are never drawn on one axis: these volumes settle from third-party sourcing, which leans heavily toward open-weight models, so a share taken across both classes would mostly describe that source rather than the market. Within a class that skew sits in every lab’s denominator alike and cancels. Frontier Token Volume and Open vs Closed Share are fixed universes — a selector would make their titles wrong.

Chart	Universe	What it answers
Token Share	Labs you tick, within one class	Each selected lab as a share of the labs selected beside it. Ticking another lab re-bases the ones already on the chart, so the drawn lines always sum to 100%. Closed and open-weight labs never share a denominator.
Volume Growth	Labs you tick	Each plotted lab indexed to 100 on the first day they all reported. The shared base moves with the selection; a lab that never settled in the window is struck through and left out of the base.
Frontier Token Volume	Closed three	The closed labs added together, indexed to 100 on the first date any of them reported. Shares cannot all rise; this is the chart that can.
Open vs Closed Share	All 11 labs	Proprietary vs open-weight as two lines on a pinned 0–100 axis, starting the day open-weight coverage begins. An open-weight model from a closed lab counts for neither side.
Input vs Output Share	One lab at a time	Prompt-side vs completion-side tokens as two mirrored lines summing to 100%. The crossing — the day output overtakes input — is the read; a within-lab ratio, so labs are never overlaid.

Share is a per-day ratio, valid on every day the selected labs reported, so it is not trimmed to the shared growth base. Growth is: drawing days before the common start would put some labs below 100 while a late arrival had no line at all.

Tracked labs

The same 11 as OTPI:

Closed. anthropic · openai · google

Open-weight. deepseek · minimax · xiaomi · qwen · moonshotai · z-ai · mistralai · meta-llama

The REST series

GET /api/token-volume publishes the closed-lab slice of the same volumes:

• growthIndex — each lab indexed to 100 at its own first settled day, not a shared day. Fixed-base: overlapping windows return identical values for the same date.
• share — of the closed-lab daily total.
• totalGrowthIndex — the three labs combined, indexed to 100 at the first date any of them reported.

startDate and endDate are required (YYYY-MM-DD). A lab that settled no volume on a day is left out of that day’s labs array rather than reported as zero. There is no lab filter: share is defined over the closed-lab total, so a single-lab read would answer share: 1 on every row.

The platform charts recompute share, the shared growth base, and the open/closed split client-side from the same settled rows. Those views are not extra fields on this endpoint. commonBaselineDate is absent from the REST response — a shared base is not a concept on the own-day series.

Related

• The Token Price Index — the per-token prices these volumes weight.
• The Workload Cost Index — the OTPI companion that prices the request rather than the token.
• Get the Token Volume Index by day — the REST schema and code samples.