Skip to main content
Morse logo

Morse

Private gateway for open-weight models with per-request privacy tiers

Morse is a privacy-focused inference gateway that serves open-weight chat models through one OpenAI-compatible API. Each request or API key sets a privacy tier (Private, TEE or E2EE), and Morse routes only to upstream providers that meet it, refusing a request it cannot serve at that tier rather than sending it lower. The same gateway runs the Morse chat app.

At a glance · prices collected Nov 19, 2026

What's good about Morse

  • Privacy tier enforced as a floor: a request is refused rather than downgraded
  • One OpenAI-compatible endpoint for open-weight models from many labs
  • Per-model input, cached-input and output rates published, with a public catalog feed
  • Pay-as-you-go with no subscription for API use

Morse LLM API prices

Pay-per-token pricing, per 1M tokens.

Prices last updated November 19, 2026

ModelInput/1MOutput/1M
$0.025$0.050
$0.036$0.037
$0.037$0.175
$0.046$0.212
$0.063$0.125
$0.063$0.188
$0.075$0.225
$0.075$0.200
$0.075$0.313
$0.075$0.225

Billing & terms

Per token
Input, cached input and output tokens billed per 1M; a model served on more than one tier can have a different rate per tier
Savings credit
Part of what a request saves through caching or routing is credited back to the balance after each UTC hour
Team extras
Dedicated enclaves and audit reports for teams
Support
Documentation and email support

Watch out for

  • Open-weight models only on the API
  • Morse charges a markup over the upstream providers' wholesale rates
  • Several models are offered on the Private tier only
  • TEE and E2EE tiers are priced higher than Private on some models

Features

Privacy tiers
Private (the provider's own hardware under zero data retention), TEE (attested confidential-computing hardware) and E2EE (end-to-end encrypted to an attested enclave), chosen per request or per API key as a floor
OpenAI-compatible API
Chat Completions endpoint at inference.trymorse.ai; switching from another provider means changing the base URL, key and model
Routing across vetted providers
Requests go to an upstream provider that meets the requested tier; the subprocessor list names the inference providers used
Prompt caching
Repeated context can be served from cache at a per-model cached-input rate
Enclave attestation
The gateway enclave is open source, and a published verifier checks that the running enclave matches the published code

Company

Published by Morse to identify the company behind this listing.

Legal entity
Watermark Labs, Inc.
Headquarters
🇺🇸 San Francisco, United States

Getting started

  1. Step 1

    Create an API key

    Sign in at console.trymorse.ai and create a key, optionally with a minimum privacy tier

  2. Step 2

    Point an OpenAI client at Morse

    Set the base URL to inference.trymorse.ai/v1 and pass the key as a Bearer token

  3. Step 3

    Pick a model and tier

    Choose a model from the models page and the tier each request needs

Where this page's facts come from

LLM API prices
  • trymorse.ai · last collected Nov 19, 2026
Profile
Reviewed Nov 19, 2026
Method
How we collect and check prices

Frequently Asked Questions

Which models does Morse serve?

Check the pricing table above for the models Morse currently serves and their per-token rates.

How do I get started with Morse?

Create an API key, Point an OpenAI client at Morse, Pick a model and tier

What are Morse's main advantages?

Morse's main advantages include: Privacy tier enforced as a floor: a request is refused rather than downgraded, One OpenAI-compatible endpoint for open-weight models from many labs, Per-model input, cached-input and output rates published, with a public catalog feed, Pay-as-you-go with no subscription for API use.

What are Morse's limitations?

Morse's main limitations include: Open-weight models only on the API, Morse charges a markup over the upstream providers' wholesale rates, Several models are offered on the Private tier only, TEE and E2EE tiers are priced higher than Private on some models.

The weekly price report

One email every Monday with the week's notable GPU and LLM API price moves, plus occasional product news. No spam.