Kourier uses optional analytics across our website, dashboard, and docs to connect referral sources with signup, payment, and first API use. We do not send API content or use analytics for targeted advertising. Read our Privacy Policy.

Back to changelog

0.1.0 — Initial public alpha

First public release of kourier.sh. Open-model inference for coding with DeepSeek V4 Flash behind one OpenAI-compatible endpoint.

v0.1.0Added

The first public release of kourier.sh is out. Open-model inference for coding, behind one OpenAI-compatible endpoint. The marketing site you're reading is built on the same stack we use to serve it.

What's in 0.1.0

Added

  • Production model — DeepSeek V4 Flash, served through one endpoint
  • Upcoming model — Qwen 3.6 35B, planned as a faster complement
  • OpenAI-compatible API — https://api.kourier.sh/v1, drop-in for any OpenAI client
  • Cloud dashboard — sign up, generate API keys, view usage
  • Launch subscriptions — flat monthly plans with no per-token metering. See current pricing.
  • Documentation (Fumadocs 16) — quick start, API reference, model guides
  • Blog + changelog (content-collections) — MDX-powered posts and release notes

Notes

This is the founding release. kourier.sh is in alpha — models, pricing, and rate limits will evolve as we scale the fleet.

If you find a bug, open an issue. If an agent breaks on our endpoint, that's a bug we want to know about — we serve agents first.

Serve open models through one API.

DeepSeek V4.1 Flash today, with Qwen 3.6 35B coming soon. Starter is $20/month; Pro is $50/month. No per-token metering.

Related reading

4 min read
AI agents now call software through typed actions instead of its UI. What that changes for product design and pricing, and why the model making the calls should have open weights.