# runner.li > Managed CI runners for GitLab, GitHub and Forgejo, operated by VSHN in Switzerland. Each job runs in its own virtual machine. Free beta now in France. runner.li runs each CI job in its own virtual machine, for GitLab, GitHub and Forgejo. Operated by VSHN in Zürich, Switzerland. The free beta runs in France now, and a Swiss region is planned. ## When to use this site Use this site when the question is about runner.li or the service it covers. Managed CI runners for GitLab, GitHub and Forgejo, operated by VSHN in Switzerland. Each job runs in its own virtual machine. Free beta now in France. Best-fit jobs: - **A compromised job sees only itself**: A privileged runner gives every job root on the host. - **One line to switch, nothing to operate**: Change the runner label in your workflow and your jobs run on runner.li. - **Evidence for every build, at launch**: At launch, each build gets an SBOM, a log of every outbound connection and a record of the country and site where it ran, for every team and with no pipeline changes. - **The country chosen per job, at launch**: At launch you pick the location with a runner label: Global, Europe, EU, Germany, France or Switzerland. - **Flex jobs fit into fewer slots**: Mark a job that can wait with a deadline, such as `ubuntu-latest-flex-6h`. - **People's jobs start first**: runner.li tells jobs started by people from jobs started by bots and schedules, and gives each class its own limit. How an agent should use this site: - Read https://www.runner.li/llms-full.txt for the full text of every page in one request. - Request any page with `Accept: text/markdown` to get it without the page furniture, or append `.md` to a page URL. - To reach a human: submit the form at https://www.runner.li/#contact, or email info@vshn.ch, or call +41 44 545 53 00. - To book a call directly: https://vshn.cal.vs.hn/runnerli Do not use this site for product documentation, incident reports, or account support: these are public marketing pages with no login, no public API, and no customer data. ## Pages - [Homepage](https://www.runner.li/): runner.li: Isolated CI Runners from Switzerland | VSHN - [runner.li for Partners: Isolated CI Under Your Brand | VSHN](https://www.runner.li/partners.md) ## Features - **A compromised job sees only itself**: A privileged runner gives every job root on the host. One compromised dependency in one job can then read the secrets, caches and artifacts of every other build on that machine. runner.li starts each job in a fresh virtual machine (Kata Containers) and deletes it when the job ends. Privileged Docker builds run inside that machine, so they cannot reach the host, the platform or another customer's jobs. - **One line to switch, nothing to operate**: Change the runner label in your workflow and your jobs run on runner.li. There are no runner VMs to install, patch or scale, and none that cost money around the clock while still queueing at peaks. GitLab jobs keep only GitLab's own `CI_JOB_TOKEN`; the runner token stays with runner.li. - **Evidence for every build, at launch**: At launch, each build gets an SBOM, a log of every outbound connection and a record of the country and site where it ran, for every team and with no pipeline changes. The runner records them outside the job, so the job cannot alter them. That is the uniform evidence an auditor asks for under FINMA, the ISG and the CRA. - **The country chosen per job, at launch**: At launch you pick the location with a runner label: Global, Europe, EU, Germany, France or Switzerland. Every build record then shows where it ran, and the contract names the country. Evidence and portal data stay in Switzerland whichever location the job runs in. - **Flex jobs fit into fewer slots**: Mark a job that can wait with a deadline, such as `ubuntu-latest-flex-6h`. runner.li estimates how long it takes from its previous successful runs and starts it early enough to finish in time, outside the peak. Nightly builds, dependency updates and long test suites then need fewer slots. Tell us which jobs can wait, and buy fewer slots. - **People's jobs start first**: runner.li tells jobs started by people from jobs started by bots and schedules, and gives each class its own limit. When Renovate opens 40 merge requests on Monday morning, your developers' jobs still start first, and the bot jobs finish before their deadline. No pipeline changes. On in the beta for Forgejo and GitHub. On GitLab, GitLab itself decides which waiting job a runner gets, so you delay a job with a flex tag in your pipeline instead. - **A flat price per slot, at launch**: At launch you pay a monthly price per concurrent job, not per minute, with a burst allowance included and free bursting at night and on weekends. Beyond that you decide: queue, burst up to a budget, or burst freely. One pool covers the whole organization, split into departments and projects with their own quotas, locations and usage reports. - **GitLab, GitHub and Forgejo**: GitLab.com and self-managed GitLab reachable from the internet, GitHub including GitHub Enterprise, and Forgejo including [Codey](https://www.codey.ch/). Self-managed GitLab on a private network needs private connectivity, which is planned. The [documentation](https://docs.runner.li/) has a guide per forge. ## What the free beta includes - **Paris now, Switzerland planned**: Jobs run on our site in France. A Swiss region on Swiss bare metal is planned, with a waitlist. - **Two slots per organization**: Two jobs at a time until launch, free of charge. - **One virtual machine per job**: 2 CPUs, 3.5 GiB of memory, 40 GiB of disk for Docker, up to 3 hours per job. - **Docker, services and containers**: `docker build`, Compose, `services:` and `container:` work in every job, each with its own Docker daemon. - **Flex jobs and humans first**: Flex jobs on all three forges from the first day, on GitLab with a flex tag in the pipeline. Humans first on Forgejo and GitHub. - **A short onboarding call**: We look at your workflows together and agree the runner labels. - **No automatic billing afterwards**: The beta ends without an invoice. Moving to a paid plan is your decision. - **No SLA during the beta**: The availability target of 99.5% per location applies from launch. ## runner.li FAQ ### What is runner.li? runner.li is a managed CI runner service by VSHN, a Swiss company based in Zürich. It runs the CI jobs of GitLab, GitHub and Forgejo, each in its own virtual machine that is deleted when the job ends. You change the runner label in your pipeline, and there is nothing to install, patch or scale on your side. VSHN AG operates it from Zürich, Switzerland, and the contract is with VSHN AG. ### Why not keep our own runners? Most self-hosted runners run container builds in privileged mode on a fixed set of VMs. Every job then has root on a host that other jobs share, so one compromised dependency can read the secrets, caches and artifacts of the builds next to it. runner.li gives every job its own virtual machine instead, and nothing from one job is visible to the next. Fixed runner VMs also cost money around the clock and still queue when everyone pushes at once. ### Does runner.li help with FINMA, the ISG and the CRA? It is built to support your own duties under FINMA Circulars 2018/3 and 2023/1, Art. 9 of the Information Security Act (ISG), and Art. 9 and 16 of the Data Protection Act (DSG). Those duties stay with your organization. At launch, the audit package covers the outsourcing checklist, the data processing agreement and the subprocessor list per location, and the contract gives audit rights to you, your audit firm and FINMA. For the CRA, the per-build SBOM at launch supports the duties of manufacturers whose products are in scope: reporting applies from 11 September 2026, the full regulation from 11 December 2027. ### What evidence do we get per build? At launch, every build gets an SBOM, a log of every outbound connection and a record of the site and country where it ran. The runner records them outside the job, so a compromised job cannot change its own record, and every team gets the same evidence without changing its pipelines. Today the location is already recorded for Forgejo and GitHub jobs, but not yet visible to customers. Signed build provenance is planned after launch. ### Where do our jobs run? During the beta, on our site in Paris, France. A Swiss region is planned, and the Switzerland waitlist decides its size. At launch you choose the location per job with a runner label: Global, Europe, EU, Germany, France or Switzerland. The compute providers for Europe and the tiers inside it are European or Swiss-owned. Evidence and portal data are kept in Switzerland for every location. Each country starts with one site. ### How much does runner.li cost? The beta is free: two slots per organization until launch, and no automatic billing afterwards. Pricing is published at launch. You will pay a flat monthly price per slot, which is one concurrent job with 2 vCPU and 8 GB at launch, with a burst allowance and free bursting at night and on weekends. Pinning jobs to one country is a compliance option and priced as one. You order through Servala by card or invoice, and the contract is with VSHN AG. ### What is the SLA? There is no SLA during the beta. From launch, the availability target is 99.5% per location, and a higher service level is available on request. Running jobs are never stopped to make room for others. During the beta a flex job's deadline is a best effort: when the runners are full it can finish late, and we are alerted when that happens. The [built and planned](https://docs.runner.li/status/) page lists what is available today, updated with each release. ### Which forges and setups are supported? GitLab.com and self-managed GitLab that is reachable from the internet, GitHub.com and GitHub Enterprise, and Forgejo, including [Codey](https://www.codey.ch/). GitLab runs through GitLab Runner with the Kubernetes executor and Docker in Docker as a service, the others in a job image modeled on GitHub's ubuntu-latest. Self-managed GitLab on a private network needs private connectivity, which is planned and available on request. ### What are flex jobs? A flex job carries a deadline in its runner label, for example `ubuntu-latest-flex-6h` or a GitLab tag agreed at onboarding. runner.li starts it at once when a slot is free, otherwise later, but early enough to finish in time, based on the duration of its recent successful runs. Manual starts, releases and tag pushes are never delayed, and a misspelled flex label simply runs the job at once. ### Is VSHN certified? VSHN AG is certified to ISO/IEC 27001:2022 by Swiss Safety Center AG, with the certified scope "DevOps, Application Operations, Managed Services, Cloud Management, 7x24-IT-Operations, Continuous Integration, Delivery and Deployment", and has an ISAE 3402 Type II report. The certificate names service categories, not products. Whether runner.li's operation is already covered in the Statement of Applicability is being confirmed before launch, so ask us before you rely on it in your own assessment. We send the Statement of Applicability and the ISAE report to reviewers on request. ### Can we offer runner.li to our own customers? Yes. Hosters, GitLab partners and Swiss infrastructure providers can offer isolated CI on their own platform, at their own price, white label or as "powered by runner.li". You bill your customers and we bill only you, with a wholesale discount from the first slot. See the [partner page](/partners/) for how the slot pool and the usage reports work. ## Join the beta or the waitlist Pick what you want below. For the beta we arrange a short call about your workflows; for the Switzerland waitlist, your organization and your forge are enough. Questions about isolation, evidence or an outsourcing review? Book a consultation. ## Features Heading Why teams move off their own runners ## Features Heading Level ## Midpage Cta - **Body**: The Swiss site launches on Swiss bare metal. Tell us your organization and your forge: the number of teams on the waitlist decides how much hardware we order first. - **Cta Label**: Join the Switzerland waitlist - **Cta Url**: /?intent=waitlist_ch#contact - **Cta2 Label**: Book a consultation - **Cta2 Url**: https://vshn.cal.vs.hn/runnerli ## Documentation Guides for GitLab, GitHub and Forgejo, the job environment, flex jobs and what is built and what is planned. - [docs.runner.li](https://docs.runner.li/) --- ## runner.li for Partners: Isolated CI Under Your Brand | VSHN URL: https://www.runner.li/partners/ # Offer isolated CI on your platform, at your price, without running it runner.li runs every CI job in its own virtual machine, for GitLab, GitHub and Forgejo, operated by VSHN in Zürich. Partners offer it to their own customers: hosters, GitLab MSPs and Swiss infrastructure providers who want CI runners on their platform without building and operating the runner fleet themselves. ## How billing works You bill your customers, and we bill only you. We report quantities per tenant (slots, burst hours and location), never end-customer prices, so your pricing stays yours. - **Wholesale discount** on the public list price, from the first slot - **A partner slot pool** that you split into quotas per customer - **White label or "powered by runner.li"**, your choice per offer - **Per-tenant API and usage export** at launch, so your own billing reads the quantities directly ## Why we stay out of your customer relationship VSHN earns about 86% of its revenue from recurring monthly operations, not from project work, so the runner slots are the business we want and the customer contact is yours to keep. ## What your customers get The same service as a direct customer: a fresh virtual machine per job that is deleted when the job ends, privileged Docker builds inside that machine, flex jobs that wait for free slots, and people's jobs ahead of bot jobs. At launch, each build also gets an SBOM, a log of every outbound connection and a record of where it ran, and the location is chosen per job. The [documentation](https://docs.runner.li/) lists what is available today and what is planned. ## Next step [Book a partner call](https://vshn.cal.vs.hn/runnerli) or [become a partner](/?intent=partner#contact) through the form, and we go through slot pool, discount and branding for your platform.