# Every GitHub Actions runner provider, priced, graded and dated

10 platforms, each read from their own pricing pages and dated.
Last full review: 2026-08-30. Dataset: https://runnerbench.com/runners.json (CC BY 4.0).

## Report card

Grades are arithmetic on the published record. Rubric weights: Cost 35%, Performance 25%, VM support 20%, Build reliability 20%.

| Provider | Overall | Cost | Performance | VM support | Build reliability |
|---|---|---|---|---|---|
| Blacksmith | A (3.22) | A | A+ | A+ | C- |
| Latchkey | A (3.15) | A+ | B | C | A+ |
| Namespace | B+ (2.85) | A | B | A+ | C- |
| WarpBuild | B+ (2.85) | A | B | A+ | C- |
| Depot | B+ (2.74) | A | A+ | C | C- |
| GitHub-hosted | B+ (2.63) | B | B+ | A+ | C- |
| BuildJet | B (2.52) | A | B | B | C- |
| StarSling | B (2.37) | A | C | C | B |
| Semaphore | B- (2.27) | C+ | B | A | C- |
| Buildkite | C (1.73) | C- | C | B | C- |

## Price at 2 vCPU, Linux x64, USD per minute

GitHub-hosted baseline: 0.006.

| Provider | 2 vCPU | RAM | Billing | Free tier |
|---|---|---|---|---|
| Namespace | 0.002 | 4 GB | per-unit-minute | Developer plan is $0, pay-as-you-go. |
| Latchkey | 0.0025 | 8 GB | per-minute | 2,000 min/mo (Developer), 4,000 (Launch), 6,000 (Scale) |
| Blacksmith | 0.004 | not published | per-minute | 3,000 min/mo |
| Depot | 0.004 | not published | per-minute | Bundled into plans: Developer $20/mo includes 2,000 Actions minutes and 25 GB cache. |
| BuildJet | 0.004 | 8 GB | per-minute | One-time $5 credit. No recurring free minutes. |
| WarpBuild | 0.004 | 8 GB | per-minute | none published |
| StarSling | 0.004 | 8 GB | per-minute | One-time 2,000 free minutes in the first month. |
| GitHub-hosted | 0.006 | 8 GB | per-minute-rounded-up | 2,000 min/mo (Free), 3,000 (Pro/Team), 50,000 (Enterprise Cloud). Private repos only. |
| Semaphore | 0.0075 | not published | per-minute | $15 credits/mo |
| Buildkite | 0.008 | 4 GB | per-vcpu-minute | Free plan: 5 users, 10 concurrent jobs, 2,000 Linux vCPU-min/mo |

## Every provider

### GitHub-hosted

The default. Free and unlimited on public repositories; metered on private ones. Verified 2026-08-30: https://docs.github.com/en/billing/reference/actions-minute-multipliers

- Larger runners are billed even on public repositories and even while plan minutes remain.
- Private repositories receive half the vCPU and RAM that public repositories do.

### Blacksmith

Bare-metal gaming CPUs in Firecracker microVMs, optimised for single-core speed. Verified 2026-08-30: https://www.blacksmith.sh/pricing

- Cites a single-thread PassMark of 4484, the highest among managed providers per the RunsOn benchmark.
- Publishes sub-3-second job boot into an ephemeral Firecracker microVM.

### Depot

Single-tenant EC2 per job with a RAM disk and an uncapped cache tier. Verified 2026-08-30: https://depot.dev/pricing

- Bills per second rather than rounding up to the minute, which matters most on wide matrices of short jobs.
- Cites 3x faster disk access from RAM disks mounted by default.

### BuildJet

Long-running drop-in replacement with a larger cache allowance than GitHub-hosted. Verified 2026-08-30: https://buildjet.com/for-github-actions/pricing

- ARM shells ship notably less RAM than AMD at the same price: 3 GB vs 8 GB at 2 vCPU.
- Default concurrency 64 AMD vCPUs / 32 ARM vCPUs; +100 vCPUs costs $300/mo.

### Namespace

Bills unit-minutes rather than runner-minutes. One unit is 1 vCPU plus 2 GB. Verified 2026-08-30: https://namespace.so/pricing

- Rates shown are prepaid at $0.001/unit-min. Overage is $0.0015/unit-min, 50% higher.
- Platform multipliers: Linux 1x, Windows 2x, macOS 10x, Linux on Apple silicon 7x.
- RAM is fixed at 2 GB per vCPU, so a memory-hungry job must buy cores to get memory. The 2 vCPU shape with 8 GB is a 4-unit shape at $0.004/min, twice the 4 GB one.
- The 64 vCPU Linux shape bills 80 units, not 64, so it costs $0.08/min rather than the $0.064 the rest of the ladder implies.

### WarpBuild

Managed runners plus a bring-your-own-cloud mode that runs inside your own account. Verified 2026-08-30: https://www.warpbuild.com/pricing

- BYOC mode is $0.002/min for the control plane; compute is billed to you by your own cloud provider.
- arm64 is priced 25% below x64 at every published size.

### Semaphore

A full CI platform rather than a drop-in runner. Adopting it means rewriting pipelines. Verified 2026-08-30: https://semaphore.io/pricing

- Self-hosted agents are $0.0025/min on any machine you supply.
- x64 is more expensive per minute than GitHub-hosted; arm64 is substantially cheaper.
- 20 concurrent jobs.

### Buildkite

Hybrid CI platform. Bills per vCPU-minute plus per-seat and per-agent fees. Verified 2026-08-30: https://buildkite.com/pricing

- Linux hosted agents are $0.004 per vCPU-minute, so a 4 vCPU agent is $0.016/min.
- Seat and agent fees usually dominate the compute comparison for teams above about ten people.
- Self-hosted agents on your own infrastructure are supported.
- macOS hosted agents are published at $0.02 per vCPU-minute. No per-runner macOS rate is published, so no figure is shown above.

### Latchkey

Managed runners that detect, repair and retry transient failures on the runner. Verified 2026-08-30: https://latchkey.dev/pricing

- The only provider in this table that attempts automatic recovery from transient failures.
- Publishes ~10s cold start and ~5s warm-pool pickup.

### StarSling

Managed runners with AI agents that open pull requests to optimize your pipeline. Verified 2026-08-30: https://starsling.dev/

- Queue time is never billed, per the published FAQ.
- Ships AI agents that open pull requests against your repo; the published examples are pipeline optimizations (caching, test sharding, parallelization) rather than repairs of failed jobs.

## Method

Every figure is read from the vendor's own public pricing page. A script re-checks each one
on review, asking a single question: is that number still present at its source?
Nothing is estimated and nothing auto-rewrites; a human resolves every flagged change.
Full methodology: https://runnerbench.com/#methodology