Updated 28 Oct 2026 · 15 models · 17 providers

TokenRate

What frontier AI models cost through cheaper routes, measured against the vendor’s own list price.

0.1%of list price: the cheapest route found, on 4 models
2.5%of list price for the median model, on its cheapest route
417InferHub routes on 28 Oct: 3 new, 15 gone

Intelligence vs price

The Artificial Analysis index against what each model costs, from its cheapest route to its list price.

Routes

Every model’s price as a share of its list price, cheapest first.

Providers

Every channel the notes name: what it costs against list price, whether it keeps your data, and how sure the notes are.

Method

How each number on this page is made.

Price

The cheapest ask per route, per 1M tokens, blended 3 parts input to 1 part output. Share of list divides it by the vendor’s own list price, blended the same way.

Intelligence

The Artificial Analysis Intelligence Index, read off its chart, so allow ±1 either way.

Confidence

How sure a provider fact is, in five steps:

ConfirmedStrongEstimatedSpeculativeUnknown

Every source has a shelf life: 3 days for an API, 14 for community notes, 30 for official pages and benchmarks. A fact past it drops one step.

Prices
InferHub GET /v1/models, cheapest ask per route, 28 Oct
Intelligence
Artificial Analysis Intelligence Index, read off the 27 Oct chart (±1)
Notes
Provider community chat, 26–27 Oct (338 messages)

Changes

What moved between the saved InferHub snapshots. Nothing here is edited by hand.

  1. 28 Oct 2026
    InferHub prices and routesCurrent

    417 routes listed, 429 the day before: 3 new, 15 gone, 112 of 214 routes priced on both days changed price.

    See the routes
  2. 27 Oct 2026
    InferHub prices and routesPrevious

    429 routes listed. Kept as the baseline for the entry above.

  3. Mar 2026
    Plan limits, from the seed fileSeed

    62 of 62 facts are past their shelf life (oldest seen Mar 2026), so each is shown one confidence step down.