Avrea vs Depot

Features at a glance

Both providers cover GitHub Actions runners and build caching. Avrea also offers tested CI fixes through Avrea Quality. Depot’s services include remote Docker builders and shared caches for local development.

Runner pricing

Depot’s $20/month Developer plan is limited to one user. On its published plans, more than one Depot user requires the $200/month Startup plan or a custom Business plan. Avrea’s pay-as-you-go plan has no monthly base fee for teams. Depot plans, Avrea pricing.

These rates exclude included minutes, discounts and plan fees. Avrea costs one-third less per x64 minute at these sizes. The ARM runners use different hardware: Apple M5 Max for Avrea and Graviton4 for Depot. Avrea pricing, Depot runner types.

Avrea’s 4-vCPU ARM runner is the fastest measured in RunsOn’s 14 October 2026 PassMark results, with the highest single-thread and all-core scores, based on five runs. Depot is not included in that dataset. Sono reports going from three minutes on Depot to two on Avrea, before repository-specific cache tuning.

Avrea’s pay-as-you-go plan includes 3,000 monthly minutes in 2-vCPU equivalents. Larger runners use that allowance proportionally. Avrea plan details.

Depot bundles several services into its plans:

Both also include Depot CI usage. Larger GitHub Actions runners consume more of the minute allowance. Depot plans, runner multipliers.

Depot’s fee covers the allowances above across its services. Avrea bills runner and storage usage without a monthly team subscription.

Our GitHub Actions pricing guide covers runner minutes, plan allowances and storage charges.

Runner, builder and subscription costs

A ten-minute job on the listed 2-vCPU x64 runners costs $0.04 on Avrea and $0.06 on Depot before included usage, plan fees or discounts.

For a team moving from Depot’s Startup plan, compare its $200 monthly subscription and overages with Avrea’s runner and storage charges after included usage. Depot’s subscription covers several services, so the value of its allowances depends on how much of the bundle the team uses.

Depot Docker workflows can incur both runner and remote-builder charges. Avrea runs the Docker build on the runner itself.

Docker builds

Both Avrea and Depot support persistent storage for Docker caching. Depot runs builds on remote BuildKit builders with native multi-platform support. The same build service works from your laptop and CI. Depot Container Builds.

Avrea runs Docker builds on its runners, with persistent disks that retain images, layers and builder state between jobs. It also offers a BuildKit cache endpoint for exporting and restoring layers. Both options let subsequent builds reuse work from earlier runs. Avrea Docker setup.

Moving a Docker build to Avrea

For a standard Docker build, replace the Depot build step with docker/build-push-action on an Avrea runner. Configure cache-from and cache-to with type=gha,url_v2=https://cache.avrea.com/. Use mode=max when exporting intermediate layers from multi-stage builds. Docker setup on Avrea.

Persistent-disk caching is another option. Enable persistent disks for Linux and Docker caching in the repository’s cache settings. Container-based Buildx builders need a stable builder name and keep-state: true to retain their state between jobs. The first run populates the cache; subsequent builds can reuse it. Persistent Docker cache.

Other build caches

Avrea automatically routes supported actions/cache usage to its cache. Its package proxies reuse downloads, while build-tool caches reuse valid outputs to skip repeated compilation or tasks. Avrea includes 25 GB per repository, shared across Actions, LFS, build and package caches, with additional cache storage available to purchase. Avrea caching, cache storage.

Avrea's documented build-cache endpoints are accessible inside Avrea runners. Depot Cache also works from developer laptops and other CI providers, so developers and CI can share a remote cache. Its supported tools are pre-configured on Depot runners. Both providers support tools including Bazel, Go, Gradle, Turborepo, Nx and sccache. Avrea build-cache scope, Depot Cache.

Avrea cache entries that haven't been used for more than seven days can be evicted. Depot Cache defaults to 14-day retention with no configured size cap, and charges for storage beyond your plan's allowance. Avrea storage policy, Depot retention and storage billing.

Our GitHub Actions caching guide covers cache keys and restore behavior.

Build performance and customer results

Sono's move from Depot

Sono reports that a build taking three minutes on Depot ran in two on Avrea, before repository-specific cache tuning. The team migrated in about fifteen minutes across a couple of pull requests, starting from an already optimized pipeline with parallel builds and an ARM integration suite. Read Sono’s case study.

This is a customer migration result; the case study does not specify matching runner sizes or repeat counts.

Debugging and AI-assisted fixes

Avrea includes searchable logs, CPU and memory metrics, and SSH into running jobs. Depot also provides metrics for its GitHub Actions runners. Avrea logs and metrics, Avrea included features, Avrea debugging, Depot runner metrics.

Avrea shows failure and flake trends for workflows and jobs, plus cache diagnostics for individual runs. Depot shows per-test results from JUnit reports; uploading those reports takes an additional workflow action. Avrea analytics, Avrea cache diagnostics, Depot test reports.

CI fixes and coding agents

In Depot, Sherlock can read the job's logs and help with the investigation. Avrea Quality uses logs, code changes and test history to propose a fix, runs your full test suite and opens a PR if it passes.

The Avrea CLI lets coding agents read failed-step logs, watch runs and dispatch GitHub workflows. Depot CI supports another approach: an agent can test local, uncommitted changes, read a failure and retry without pushing each attempt. Its AI error summaries also point back to the failed step. Using Depot CI involves migrating workflows; changing to Depot's GitHub Actions runners alone does not give you that workflow.

Avrea has sandboxes in beta. Depot Remote Agents run Claude Code sessions in cloud sandboxes, billed separately from runner usage.

Network access and your own AWS account

Both providers offer static IPs. Avrea lists them under its Enterprise features at $100 per IP per month. Depot provides static outbound IPs through a NAT gateway in a custom deployment on its Business plan. Avrea static IP pricing, Depot private-resource access.

Both providers let you restrict outbound network traffic. Avrea enforces its Enterprise firewall outside the runner VM, where the build cannot change the rules. Depot's Linux runner filters operate inside the runner; its documentation says a process with root access can modify them. Avrea firewall, Avrea plans, Depot egress filters.

Both allow outbound traffic by default, and some services remain reachable after you set restrictions. Avrea has exceptions for GitHub services and Azure Blob Storage. Depot has allowances for its own services, GitHub Actions and AWS. Avrea exceptions, Depot exceptions.

On its Business plan, Depot Managed can run compute and keep cache data in your own AWS account. Depot Managed.

Choosing between Avrea and Depot

Avrea has no monthly team subscription, lower x64 minute rates at the listed sizes, and the ARM CPU performance shown above. Build caching, CI diagnostics and Avrea Quality address slow builds and the work of investigating and fixing failures. Sono’s migration shows how one team used Avrea in an already optimized pipeline.

For a Docker-heavy pipeline, compare the complete build on Avrea with the current Depot runner and remote builder together. That captures the time and charges for the whole build, including cache reuse, rather than comparing the runner rates alone.

Related reading

Other comparisons: Blacksmith, Namespace, WarpBuild, Ubicloud and GitHub-hosted runners. See all comparisons. For more about Avrea, see runner performance and caching and security controls.