Koyeb
Serverless platform with scale-to-zero GPUs, now part of Mistral Compute.
Acquired — still operating
acquired by Mistral AI Feb 2026; folded into Mistral Compute
About
Koyeb is a Paris-founded serverless platform that runs apps as microVMs on bare metal across the US, EU, and Asia, with per-second billing and scale-to-zero on both CPU and GPU (up to 8x H200s). It added serverless Postgres and Koyeb Sandboxes, isolated environments aimed at AI agent workloads.
In February 2026 Mistral AI acquired Koyeb in its first acquisition, folding the platform and its 13-person team into Mistral Compute, the sovereign-European AI cloud play backed by major data center investments. The platform keeps operating, but the strategy has tilted toward enterprise and AI inference workloads.
Limitations to note: the free tier closed to new signups after the acquisition, so new accounts start on paid plans; Mistral integration direction is still unfolding; and GPU availability can vary by region.
What it can do
- Scale-to-zero serverless on CPU and GPU
- GPU inference up to 8x H200 with per-second billing
- Sandboxes for AI agent workloads
- Managed Postgres and global regions
Setup
Sign up on the vendor site and deploy via CLI, API, or git push. New accounts start on paid plans.
Prerequisites
A containerized app or code to deploy; a paid plan for new accounts.
Use cases
Works with
Trust score — 8/10
Scored 8/10 (verified tier, metrics-v1). Evidence: Verified via vendor site/docs/GitHub and reputable coverage, wave 83; Trending score 80/100 from researcher assessment.
Sources
Every fact above was checked against these sources. Where a claim comes from the provider and couldn't be independently verified, the text says so.