NEURAPHICCloud
Support Open Console

PRICING

Pay for what you use.
Nothing more.

Per-token pricing for every Neuraphic model, billed by the second. No platform fees. No seat fees. New accounts start with $100 in free credit.

The cost of running a model should be a line item, not a negotiation. Neuraphic Cloud publishes a single price per million tokens for each model and lets you read it before you commit. The Console returns a usage receipt for every request — input tokens, output tokens, cache hits, latency — so the bill at the end of the month is something you have already seen.

Every model on this page runs on infrastructure we built and operate. Praeth, Claeth, and Prion share the same Console, the same SDK, and the same auditable logs. You move between them by changing a string.

Models

Prices are per 1 million tokens. Tokens are roughly four characters of English text.

FRONTIER

Praeth

Deep reasoning, long-horizon agents, and high-stakes review.

$12.00input / 1M tokens
$60.00output / 1M tokens
  • 200K context
  • Tool use, vision, JSON mode
  • Streaming and batch

GENERAL

Claeth

The default. Fast, capable, suitable for most production workloads.

$2.50input / 1M tokens
$12.50output / 1M tokens
  • 200K context
  • Tool use, vision, JSON mode
  • Streaming and batch

FAST

Prion

Low-latency inference for high-volume, real-time paths.

$1.20input / 1M tokens
$6.00output / 1M tokens
  • 128K context
  • Tool use, JSON mode
  • Streaming and batch

EFFICIENT

Prion mini

Cheapest path. Classification, routing, and lightweight extraction.

$0.40input / 1M tokens
$2.00output / 1M tokens
  • 64K context
  • JSON mode
  • Streaming and batch

FREE TRIAL

$100 in credit, free for 90 days.

Run real workloads against every model with no commitment. The credit applies to all usage on Neuraphic Cloud. After 90 days, you only pay for what you use beyond the credit.

Start free Talk to sales

Billing, in detail

How is a token counted?
One token is roughly four characters of English text, or about three-quarters of a word. The Console shows the exact count for every request in the usage receipt.
How is output billed?
Per token generated. Streaming and non-streaming are billed identically. If a request fails before any output is produced, you are not charged.
What happens if a request times out?
You are billed for any tokens that streamed back successfully and nothing else. Hard infrastructure failures on our side are not billed.
How does the $100 free credit work?
Every new account is provisioned with $100 in credit, valid for 90 days. The credit applies automatically to any usage across any model. You do not need to add a payment method to start; you only add one when you want to continue past the credit or the 90 days, whichever comes first.
Are there minimums or seat fees?
No. Neuraphic Cloud is metered. You pay for tokens. There is no seat fee, no platform fee, and no monthly minimum.
How do invoices work?
Pay-as-you-go accounts are charged at the end of each month. Committed accounts receive a single monthly invoice with usage detail per model, per project, and per API key.
Where is data processed?
By default, requests run in the United States. Enterprise customers can pin processing to specific regions and disable retention. See the Safety page for the data handling policy.

Build on the same models we run.

Open the Console and start with the free credit.

Open Console Talk to us