Touchmark

Platform

Marketplace in which buyers reserve AI model inference as a set number of tokens over a month or as hourly throughput, specify availability, first-token latency and speed commitments, and providers bid to serve the order, with resale of unused capacity; priced by a published curve from each model's list rate, with buyers paying no transaction fees.

Find alternatives to Touchmark

TypePlatform

RoleStandalone tool

Bought forBuy data-centre AI compute, Track and cut what we spend on AI models

AvailabilitySold

DeploymentSaaS

Intended forDeveloper, Enterprise

SecurityVendor-asserted

CertificationsNot recorded

Trains on customer dataNo

Zero data retentionUnknown

Data residencyNot recorded

Security featuresNot recorded

StatusActive

Sold withinSold on its own

PricingUsage-based — official pricing ↗

Pricing, Deployment, Sold within — touchmark.ai · 20 Oct 2026 Security — app.touchmark.ai · 20 Oct 2026

What it does

Model inference
GPU cloud

Features 4

Reserved volume or throughputBuyers reserve a set number of tokens over the month or hourly throughput.
Service levelsProviders commit to how often the model answers, how fast the first token comes and how fast the rest follow.
Provider biddingMultiple providers bid on the same order.
ResaleCapacity can be used or sold to another buyer.

touchmark.ai

Sources

Pricing

Pricing page · 20 Oct 2026

Pricing page publishes the formula: a 1B-Token allocation at the model's posted per-1M output rate, discounted by up to 20 percent the further ahead delivery is bought; providers set the list rate per listing.

Description

Official documentation · 20 Oct 2026

Home page: "A new instrument for buying and selling reserved inference."

What it does

Official documentation · 20 Oct 2026

Home page: "A new instrument for buying and selling reserved inference."

Constraints