whichdevtool?
Menu
StacksDirectory Pricing changes Glossary
VPS and cloud servers App hosting and PaaS Managed databases Transactional email APIs Email marketing and newsletters Domain registrars CDN, DNS and edge security Monitoring, uptime and error tracking Auth and identity Frontend and full-stack frameworks Backend frameworks UI component libraries AI model APIs File and object storage Payments and billing Background jobs, queues and workflows ORMs and database toolkits

Serverless containers and GPUs · by Koyeb (Mistral AI)

Koyeb

Serverless platform for containers, APIs and GPU inference on bare metal in 50+ locations, with scale-to-zero and per-second billing; joining Mistral AI and closed to free signups since 2026.

Visit koyeb.com ↗ Alternatives Paid EU region
Koyeb homepage
§1

Overview

Koyeb runs your code or container images in lightweight VMs on its own bare-metal servers, with automatic HTTPS, HTTP/2 and HTTP/3, WebSockets, gRPC, private networking and a global edge. Services can autoscale horizontally and scale to zero when idle, with CPU cold starts under 250 ms according to Koyeb. It also offers a wide range of GPUs (L4 to H200 and B200) billed per second, serverless Postgres with pgvector, NVMe volumes and sandboxes for running agent code.

In February 2026 Koyeb agreed to join Mistral AI and become part of Mistral Compute. As part of that, it stopped offering the free Starter plan to new users: new organizations must subscribe to Pro ($29 a month with $10 of compute included), Scale ($299 with $100 included and a 99.9% SLA) or Enterprise (from $1,000). Existing Starter organizations keep their free instance.

Bandwidth is generous, with 1 TB included on every paid plan. The company's stated focus is now inference, sandboxes and MCP servers, so weigh the roadmap if you want a general-purpose web host.

§2

Pricing and plans

Plan Price What you get
Pro $29/mo + compute $10 compute included, 10 users, 100 services, volumes, email and chat support
Scale $299/mo + compute $100 compute included, 50 users, AWS regions, 99.9% SLA
Enterprise From $1,000/mo SSO, RBAC, audit trail, SOC 2/ISO 27001, 99.99% SLA
L4 GPU $0.70/hr 24 GB VRAM, 6 vCPU, 32 GB RAM
H100 GPU $2.50/hr 80 GB VRAM, 15 vCPU, 180 GB RAM
Postgres Small $0.04/hr 0.25 vCPU, 1 GB RAM; storage $0.50/GB-month

Compute is billed per second on top of the plan fee; CPU services start at $0.000006 per second according to the pricing FAQ, but the pricing page we read lists only GPU instance rates, so CPU instance prices are not given here. 1 TB of bandwidth is included; extra is $0.02/GB in the EU and US and $0.04/GB in Asia. The free Starter plan (one free 512 MB instance) is closed to new signups since February 2026 but kept for existing organizations. Early-stage startups can apply for up to $30k in credits.

Source: www.koyeb.com/pricing · prices as published, excluding tax.

§3

What it does

Scale to zero
Services and serverless Postgres can sleep when idle and wake on traffic.
GPU instances
From RTX 4000 SFF Ada at $0.50/hr to 8x H200 at $24/hr, billed per second.
Serverless Postgres
Managed, highly available Postgres in Washington D.C., Frankfurt or Singapore, with pgvector.
Global deployment
Pick one or many locations; Scale adds 50 on-demand regions and AWS regions.
Infrastructure as code
CLI, API, Terraform and Pulumi support.
Sandboxes
Isolated environments for executing untrusted or agent-generated code.
§4

Strengths and limitations

Strengths

  • 1 TB of bandwidth included on every plan, $0.02/GB after in the US and EU.
  • Wide GPU range with per-second billing and scale-to-zero.
  • Frankfurt region and an EU-founded company.
  • SOC 2 and ISO 27001 on Enterprise.

Limitations

  • No free plan for new users since February 2026; a card and the $29 Pro plan are required.
  • Product direction is shifting toward Mistral Compute and AI workloads.
  • SSO, RBAC and compliance certifications are Enterprise-only.
  • The public pricing page lists GPU rates but not CPU instance rates.
§5

Who it suits

Good for

  • AI inference endpoints that should scale to zero
  • Bursty APIs deployed in several regions
  • Teams that want GPUs and web services on one platform

Look elsewhere if

  • Hobby projects looking for a free tier
  • Teams that want a stable, general-purpose PaaS roadmap
§6

Pricing history

  • Koyeb closes its free Starter plan to new users

    Announced 17 Feb 2026 with Koyeb's agreement to join Mistral AI: new users must subscribe to Pro ($29/month), Scale or Enterprise; existing Starter organizations keep their free instance. Source ↗

§7

Alternatives to Koyeb

Tool From Free option Stages
Vercel Frontend and serverless platform
$20/mo Free tier
Netlify Frontend and serverless platform
$9/mo Free tier
Scalingo European managed PaaS
€7.20/mo Trial
Render Managed PaaS
$7/mo Free tier
Clever Cloud European managed PaaS
€6/mo Trial
Cloudflare Workers (incl. Pages) Edge serverless platform
$5/mo Free tier

All app hosting and PaaS →

§8

Questions

Does Koyeb still have a free tier?

Not for new users. Since the February 2026 announcement of joining Mistral AI, new signups must choose Pro ($29/mo), Scale or Enterprise; existing Starter organizations keep their free instance.

How much bandwidth does Koyeb include?

1 TB per month on every plan, then $0.02/GB in the US and EU and $0.04/GB in Asia.

Was Koyeb acquired?

Koyeb announced on 17 February 2026 that it had entered a definitive agreement to join Mistral AI, subject to closing conditions.

Sources read for this page