Pricing
Faster GitHub Actions runners. Agents that keep making them faster.
Per-minute CPU and GPU pricing, with unlimited concurrency. Optimization agents are included with paid usage.
Savings
After the optimization agents have worked on the suite.
- Cut runtime by up to
91%
The agents shard suites, fix caches and cut critical path.iWhere this figure comes fromMeasured results from customers
- Cut per-minute cost by
33%
$0.008/min against GitHub’s $0.012/min on 4 vCPU / 16 GB.
- Total savings of
94%
9% minutes × 67% rate = 6% bill.
A shorter run billed at a lower rate.
CPU Runners
Ubuntu 24.04 on 5th Gen AMD EPYC, from 2 to 64 vCPU. All prices in USD.
Scroll horizontally to compare rates and copy runner labels.
| Size | StarSling / min | GitHub / min | You save | runs-on: label |
|---|---|---|---|---|
| 2 vCPU / 8 GB | $0.004 | $0.006 | 33% | |
| 4 vCPU / 16 GBdefault | $0.008 | $0.012 | 33% | |
| 8 vCPU / 32 GB | $0.016 | $0.022 | 27% | |
| 16 vCPU / 64 GB | $0.032 | $0.042 | 24% | |
| 32 vCPU / 128 GB | $0.064 | $0.082 | 22% | |
| 64 vCPU / 256 GB | $0.128 | $0.162 | 21% | |
| CustomContact sales | Talk to us about your team's compute needs. | |||
GitHub rates are Linux x64 larger-runner list prices. View GitHub pricing.
StarSling Review Runners bill at these same CPU rates. Explore StarSling Review Runners.
Full configurations in the compute reference.
GPU Runners
Dedicated NVIDIA GPUs. Each rate includes the GPU and its host: 4 vCPU, 16 GB RAM and 100 GB disk. All prices in USD.
GitHub Actions offers one GPU runner, a 16 GB NVIDIA T4. Every comparison here is against that card.
Scroll horizontally to compare rates and copy runner labels.
| GPU | Speed (FP32 TFLOPS) | vs GitHub (T4) | StarSling / min | You save | runs-on: label |
|---|---|---|---|---|---|
| NVIDIA RTX PRO 6000default | 120 | 14.8x faster | $0.05922 | 92% | |
| NVIDIA RTX 5090 | 104.9 | 13.0x faster | $0.03022 | 96% | |
| NVIDIA RTX 4090 | 83 | 10.2x faster | $0.02522 | 95% | |
| NVIDIA H100 | Coming soon | ||||
Speed is NVIDIA's published FP32 TFLOPS, computed from its published CUDA core count and boost clock for the cards where NVIDIA states those instead. You save is for equal FP32 work: the same job takes 14.8x the minutes on GitHub's T4 as on the RTX PRO 6000, so the saving is measured against those minutes at GitHub's own rate rather than against its per-minute price. That runner is a 16 GB NVIDIA T4 at 8.1 TFLOPS and $0.052/min, on a 4 vCPU, 28 GB host. View GitHub pricing.
Sources: GitHub: T4 GPU runners, GitHub Actions larger runner specs, GitHub Actions runner pricing, NVIDIA T4, NVIDIA RTX PRO 6000, NVIDIA RTX 5090, NVIDIA RTX 4090.
Tell us which GPU you want and what you plan to run on it, and we will open access for your org.
Full configurations in the compute reference.
Estimate your monthly cost
Choose a runner and your monthly job minutes. Estimates use list prices and bill each provider for the minutes the same job takes there: 1.7x on GitHub's CPU runners, and on GPU the minutes its T4 needs to do the same FP32 work.
Monthly estimate
GitHub
$204.00
10,000 min on StarSling = 17,000 min on GitHub
StarSling
$80.00
10,000 billed min
Save $124.00
per month · 61% less
Rates from the StarSling configuration reference and GitHub pricing.
Free skills for your coding agent
Run these free, open-source skills with your own coding agent to audit and improve your GitHub Actions workflows.
/ci-speedupci-speedup samples your own Actions runs, finds the check that gates your pull requests, and hands your coding agent the evidence to shorten it.See how ci-speedup works/ci-scoreci-score reads a local checkout, grades it against the best-practice rubric we publish, and ranks a fix for every gap it finds.See how ci-score works/ci-secureci-secure scans your workflows for the ten critical attack vectors, explains each as a plain-English attacker scenario, and fixes the ones you pick.See how ci-secure works/slingsling is the agent-first CLI for StarSling Runners: why a job failed, where the time went and what CI costs, and the skill teaches your agent to use it.See how sling works