Skip to content

Foundry / by Edelux

Ship on your terms.

Build and ship apps, run agents and compute jobs, and choose where your models run. Keep your code in Git and set your own rules for data access.

Build, ship, and operate.

Start with a Git repository or build in Foundry. The same platform handles app delivery, agent workflows and the compute they need.

Build in a workspace

Use Chat and Build to turn an idea into an app, or import an existing repository into the browser IDE. Keep project files, secrets and checkpoints together.

Open Chat and Build

Ship from GitHub

Connect a repository for pull request previews and production builds. Publish with public or private access, and promote an earlier build if a release needs to roll back.

Explore projects and releases

Operate agents and workflows

Connect tools and knowledge, test agents against versioned cases, then run workflows on a schedule or a signed webhook. Pause sensitive steps for human approval.

See an approval workflow

Run the compute behind it

Use sandboxes for code, submit GPU jobs, or serve open-weight models. Choose managed inference or export the deployment files to run models on your infrastructure.

See model deployment options

Give the agent a job, not the keys.

Define which tools an agent can call, then decide which actions need a person’s signature before the workflow continues.

Open the workflow builder
Invoice reviewExample run
  1. Read the invoice

    The agent extracts the supplier, reference and amount.

  2. Look up the purchase order

    A workflow tool fetches the order and checks the amount.

  3. Wait for finance

    A person reviews the match before the workflow continues.

StatusAwaiting approval

Sovereignty is a set of decisions.

Keep the source in your Git repository. Choose which models run on your own infrastructure, who can use your data and tools, and how much a workload can spend.

  • Limit who can run what

    Assign workspace roles and issue API keys scoped to the models a tool needs.

    Team & roles
  • See what each team spends

    Set budgets and quotas, track costs by team, and get alerts before a budget runs out.

    Costs & budgets
  • Keep a verifiable record

    Review who changed what. A per-organisation hash chain lets you verify the audit trail.

    Audit trail
  • Track AI Act obligations

    Record a system's risk classification and track reporting deadlines in one place.

    Compliance workspace

For enterprises.

Get the entire Foundry platform as managed SaaS, including projects, app delivery, agents and model serving. Use your SSO, define approval rules and keep an audit trail. Run models on your own infrastructure where policy requires it.

Plan an enterprise rollout

Choose where models run.

Your hardware

Export the deployment files, read them, and run the model on Kubernetes, Docker Compose or a Linux host.

Explore bundles

Managed inference

Run a hosted CPU endpoint. You configure the model; Edelux handles the underlying compute.

Explore endpoints
  • Kubernetes
  • Docker
  • Google Cloud
  • Microsoft Azure
  • AWS
  • OVHcloud
  • Scaleway
  • Bare metal Linux

Pick the weights.

Check the licence and context window before you import a release into your registry.

55 models

Listings link to upstream weights, not hosted endpoints. Serving engines include vLLM, SGLang, TGI, Ollama.

Keep the client.

Point your tools at the OpenAI-compatible gateway and give each one its own key.

CLI setup
curl -fsSL https://edelux.studio/setup | bash

Configures Oh My Pi, Claude Code, Codex CLI, OpenCode, Edecode, Hermes Agent, Continue.dev, Aider.

Read the script
Manage API keys

A market for spare inference.

When demand outgrows your fleet, buy inference from another operator. If your GPUs have room to spare, list capacity yourself.

  • Need more compute?

    Buy inference capacity listed by other operators without provisioning another cluster.

    Browse the marketplace
  • Have capacity to spare?

    Publish an inference offer. The ledger records usage, fees and settlement against your wallet.

    Sell inference
  • Serving your own customers?

    Resell inference through your own gateway with a separate key for each client.

    Explore resale

Personal plans for solo developers.

Renew Chat and Build monthly with crypto; there’s no auto-renewal and unused credit expires. API requests use a separate prepaid balance.

Starter

€10/ month

€20 inference credit for Chat and Build

Choose Starter

Plus

€100/ month

€200 inference credit for Chat and Build

Choose Plus
Compute and marketplace rates
Marketplace fee
8% of a settled sale
Compute billing
Per second while the workload runs

Bring your organisation onto Foundry.

Tell us what your teams need to run and what your policies require. We’ll plan the full SaaS rollout with you and identify anything that should stay on your infrastructure.

Talk to enterprise sales