---
title: "StarSling vs Blacksmith: Fast Runners + AI Optimization"
description: "How StarSling compares to Blacksmith: runner OS coverage, vCPU ceiling, caching, and pricing. Both are drop-in runners; StarSling adds agents that open CI PRs."
url: https://starsling.dev/compare/blacksmith
canonicalUrl: https://starsling.dev/compare/blacksmith
---

# StarSling vs Blacksmith runners

Last updated: 2026-07-16

StarSling is AI-native CI for GitHub Actions: drop-in Ubuntu runners plus agents that open optimization PRs for your workflows.

Blacksmith provides fast, drop-in GitHub Actions runners on better hardware, with runner families beyond Linux and build-infrastructure features such as caching. StarSling is also a drop-in runner replacement, focuses on Ubuntu Linux runners up to 64 vCPU, and adds the AI-native layer: agents that analyze your workflows, logs, and telemetry and open pull requests that optimize caching, installs, parallelization, and test sharding against the [best practices StarSling agents check for](/best-practices/github-actions). The category difference is fast runners plus build infrastructure vs fast runners plus self-driving optimization.

## Head to head

| Category | StarSling | Blacksmith |
|---|---|---|
| Drop-in runner replacement | Yes, swap `ubuntu-latest` for `starsling-ubuntu-24.04`. | Yes, a drop-in GitHub Actions runner replacement via the `runs-on` label ([Blacksmith quickstart](https://docs.blacksmith.sh/introduction/quickstart), checked 2026-07-14). |
| AI optimization PRs | Yes. AI agents analyze logs/telemetry and open optimization PRs. For new accounts, AI-powered optimization PRs are only available to customers on paid plans and are not enabled by default. | No. Blacksmith's published product surface is runners and caching; its docs list no agent that opens optimization PRs (checked 2026-07-14). |
| GitHub Actions support | Runs your existing GitHub Actions workflows unchanged. | Yes. Runs existing GitHub Actions workflows. |
| Runner OS coverage | Ubuntu 24.04 Linux runners. | Linux x64 and ARM and macOS runner families in the [published instance-type list](https://docs.blacksmith.sh/blacksmith-runners/overview), plus Windows Server 2025 in public beta (checked 2026-07-14). |
| Linux runner ceiling | 64 vCPU (`starsling-ubuntu-24.04-64`). | 32 vCPU in the [published Blacksmith Linux runner table](https://docs.blacksmith.sh/blacksmith-runners/overview) (checked 2026-07-14). |
| Caching | GitHub Actions caching works unchanged; the agents open PRs that fix cache keys, cache misses, and unnecessary installs. | Docker layer caches are kept on per-Dockerfile sticky disks that persist between runs, billed per GB-month as an add-on ([Blacksmith Docker build caching](https://docs.blacksmith.sh/blacksmith-caching/docker-builds), checked 2026-07-14). |
| Primary focus | Fast runners plus self-driving AI optimization of the pipeline. | Fast drop-in runners, caching, and build infrastructure. |
| Pricing model | Usage-based per-minute, from $0.004/min (2 vCPU) to $0.128/min (64 vCPU); queue time never billed; no seat fees. | Usage-based per-minute, with the rate set by runner OS and architecture, plus per-GB add-on charges for cache storage. See [Blacksmith's pricing page](https://www.blacksmith.sh/pricing) for current rates (checked 2026-07-14). |

## How Blacksmith runners work

Blacksmith is a drop-in GitHub Actions runner replacement: you keep GitHub Actions and change the `runs-on` label to a Blacksmith runner label ([Blacksmith quickstart](https://docs.blacksmith.sh/introduction/quickstart), checked 2026-07-14). Its published instance list spans Linux x64, Linux ARM, macOS, and Windows Server 2025 beta families, with Linux instance types up to 32 vCPU ([Blacksmith runner overview](https://docs.blacksmith.sh/blacksmith-runners/overview), checked 2026-07-14).

For Docker builds, Blacksmith's docs describe per-Dockerfile sticky disks that persist Docker layer caches between runs, billed per GB-month as an add-on ([Blacksmith Docker build caching](https://docs.blacksmith.sh/blacksmith-caching/docker-builds), checked 2026-07-14). That changes the cold-cache problem: repeat builds can reuse layers instead of rebuilding everything from zero.

That is runner and cache acceleration. It can make existing work start and run faster, but it does not remove work the workflow should not be doing. StarSling's axis is the agent layer: find cache misses, redundant installs, unsafe serial work, unsharded suites, and oversized runners, then open reviewable PRs that change the workflow itself.

## Real StarSling optimization PRs on existing workflows

The difference between fast runners and self-driving CI is visible in the customer PRs. StarSling agents do not just run jobs on different hardware; they edit the workflow and hand the team a PR to review. These are published case-study windows, not head-to-head Blacksmith benchmarks; follow each case study for source method and caveats.

- [Better Auth](/customers/better-auth): agents fixed Turborepo cache configuration and cached Playwright browser installs. The published case study measures E2E at 2m 22s to 1m 04s and CI at 1m 40s to 1m 02s.
- [Mastra](/customers/mastra): agents sharded E2E kitchen-sink across parallel jobs ([PR #15888](https://github.com/mastra-ai/mastra/pull/15888)) as part of the workflow speedup; E2E measured 14m 16s to 10m 50s.
- [Partcl](/customers/partcl): on synthesis-heavy CI, the agents right-sized runner classes and cut heaviest-job compute cost 13x per run while queue p95 fell from 9.5 minutes to 35 seconds.

## Best fit

**Choose StarSling if:** Teams that want fast runners AND an AI layer that continuously improves the pipeline through reviewable PRs.

**Choose Blacksmith if:** Teams that primarily want faster drop-in runners, non-Linux runner families, or caching/build infrastructure without an optimization-agent layer.

## FAQ

### How is StarSling different from Blacksmith?

Both are fast, drop-in GitHub Actions runners. Blacksmith also covers non-Linux runner families and caching/build-infrastructure features. StarSling focuses on Ubuntu Linux runners up to 64 vCPU and adds AI agents that analyze your workflows and open optimization PRs, so the pipeline keeps getting faster, not just the hardware.

### Is StarSling a drop-in replacement like Blacksmith?

Yes. Swap your `runs-on` label to `starsling-ubuntu-24.04`; your workflows, actions, and secrets are unchanged.

## References

- [StarSling Runners are now generally available](https://starsling.dev/blog/starsling-runners-are-now-generally-available)
- [Launch YC: StarSling Runners - Self-Driving CI](https://www.ycombinator.com/launches/R7B-starsling-runners-self-driving-ci)
- [StarSling docs](https://docs.starsling.dev)
- [What is AI-native CI?](https://starsling.dev/ai-native-ci)
- [All StarSling comparisons](https://starsling.dev/compare)
- [Benchmark on your own workflows](https://docs.starsling.dev/performance/benchmarks)
- [Migration guide](https://docs.starsling.dev/configuration/migration-guide)
- [Runner label reference](https://docs.starsling.dev/configuration/label-reference)
- [Runner instance types](https://docs.starsling.dev/runners/instance-types)