Skip to content
New York Token Exchange

Token price index · $NYTE

NYTE Token Index

AI inference prices, in US dollars per billion tokens.

$376per billion tokens-5.7% over 7 daysAs of
Line chart of the NYTE Token Index in US dollars per billion tokens from October 2025 to September 2026, peaking near $918 in January and ending near $376.Line chart of the NYTE Token Index in US dollars per billion tokens from October 2025 to September 2026, peaking near $918 in January and ending near $376.

Four grades

Priced by capability.

AAA Frontier$2,621-13.3% (7D)
A Advanced$1,122-7.3% (7D)
B Professional$206+3.2% (7D)
C Efficient$400.0% (7D)
Log-scale chart of the four NYTE grade indices in US dollars per billion tokens from October 2025 to September 2026, ending with Frontier near $2,621, Advanced near $1,122, Professional near $206, Efficient near $40.Log-scale chart of the four NYTE grade indices in US dollars per billion tokens from October 2025 to September 2026, ending with Frontier near $2,621, Advanced near $1,122, Professional near $206, Efficient near $40.

AAA Frontier

Frontier research, the hardest reasoning, strategic synthesis. The leading labs' current flagships.

Example models

  • Claude Opus 5.5
  • GPT-6 Astra
  • Claude Fable 5.1

A Advanced

Coding agents, complex analysis, professional workflows that need near-frontier capability.

Example models

  • Kimi K3
  • Claude Opus 5
  • GLM-5.3
  • GPT-6 Sol
  • Claude Fable 5

B Professional

Production assistants, structured analysis, code review, document work at scale.

Example models

  • DeepSeek V4.1 Flash
  • GLM-5.3 Flash
  • DeepSeek V4 Pro
  • GLM-5.2
  • Gemini 3.8 Flash

C Efficient

Extraction, summarization, classification, routing, high-volume support.

Example models

  • DeepSeek V4 Flash
  • Gemini 2.5 Flash-Lite
  • GPT-5 mini
  • GPT-4.1 mini

The same published index, capability grades and chart data as TrustedRouter's Token Index. Prices are in US dollars per billion tokens. The date above identifies this published snapshot.

Provisional. Not for trading. Index levels are a reference, not executable provider quotes or a guarantee of available capacity.

For GPU operators and neoclouds

List spare inference.
Keep your own brand.

Turn spare GPU capacity into model inference at your own published rates. Offer a white-label service under your provider identity, with TrustedRouter handling customer access, routing and billing. Provider approval, rates and payout terms are agreed during onboarding.

List spare capacityExplore the open-source inference stack

Confidential routes require verified hardware and serving software, independent attestation and explicit zero retention.