The CI Challenge: save 10% or get $1,000
Technical report

High performance sandbox benchmarks

Compare 15 sandbox providers on one target hardware shape, across real developer, CI/CD, and reinforcement-learning environment workloads.

Commit
8d58408dc4ca
Run date
2026-11-16
Target spec
4 vCPU · 8 GB · 40 GB
Replicates
n = 1 to 12 sandboxes / case
Providers
15 validated

We measured the end-to-end time developers and agents experience across environments from 15 sandbox providers, including Firecracker- and libkrun-based stacks, as they complete software engineering tasks. Each repository run starts with a clone and cold dependency install, then continues through that repository's own lint, typecheck, build, and test tasks.

  • Partial results: 689 of 1008 planned cells complete; 319 incomplete; 0 excluded. Provider coverage is uneven, so this run does not establish a complete comparison.
  • Real-workload suites: 3 repositories, each run cold on 12 replicate sandboxes through that repository's own CI tasks: clone, install dependencies, lint, typecheck, build, test. The repositories are Better Auth, Mastra, and OpenClaw.
  • Synthetic suites: 10 versioned Phoronix Test Suite profiles. Of those, 8 are industry-standard profiles published through OpenBenchmarking.org. Each suite runs on 3 replicate sandboxes (5 for CPU) and repeats until Phoronix's own convergence criterion is met. Methodology lists what every suite measures.

Explore any metric · scan the full table.

Results explorer

Choose a statistic (min, p50, p95, or max), then a real-world pipeline or synthetic benchmark. Bars rank providers best-first against the leader. Hover or focus a row for instance metadata and the raw sample distribution; select it to jump to that metric in the table below.

starsling.dev/hpc-sandbox-benchmarks

High performance sandbox benchmarks

Real-world pipelines

s · lower is better · p50

Every environment · target 4 vCPU · 8 GB · 40 GB

Not reported: Baseten Sandboxes, Daytona (container), Tangle Sandbox.

Real world · Better Auth · Cold · total (Σ task medians) · Metric p50

light to dark = execution order

Metric
Real
Synthetic
Pick a synthetic benchmark family to chart a metric

What these sandbox benchmarks answer

Findings from run 43863642458, tested 2026-11-16, targeting 4 vCPU · 8 GB · 40 GB.

Key findings

Firecracker vs libkrun

On Node.js Web Tooling, the highest measured Firecracker result is Daytona (VM): 21.73 runs/s (p50, 5 sandboxes); the highest measured libkrun result is microsandbox: 16.77 runs/s (p50, 5 sandboxes). Hardware, storage, and region vary between these environments. This measures the deployed stacks, not the performance cost of either runtime in isolation.

Cold start vs workload performance

Daytona (VM) has the shortest Better Auth cold dependency install: 7.85 s at p50. Modal (VM) has the shortest full workflow: 116.4 s, summing task medians. The install leader is not the workflow leader. Cold install starts inside a running sandbox; sandbox creation-to-ready time was not measured.

CPU, RAM, disk, and network tradeoffs

Namespace leads Node.js Web Tooling at 23.43 runs/s; Daytona (VM) leads STREAM Triad at 260,182 MB/s. A CPU result does not predict memory bandwidth. Use the disk and WAN rows for their specific I/O patterns; neither a 4 KB random-write test nor transfer throughput is a general application-speed score.

Which dimensions matter for agents and CI?

For a fresh agent sandbox, compare setup tasks such as clone and dependency install. For a complete CI run, compare the repository pipeline closest to yours in the Results explorer. Synthetic CPU, memory, disk, and network tests are separate measurements; this run does not establish which resource caused a task to take longer.

Not measured in this run: sandbox creation-to-ready time, runtime RAM overhead, or network round-trip latency. STREAM measures memory bandwidth; iperf3 measures throughput. These results cannot establish which runtime boots fastest or uses the least RAM.

Measured summary

Highest throughput or shortest duration among measured environments, selected separately for each row. This is not an overall provider ranking. Non-total measurements and pipeline task values are p50; workflow totals sum per-task p50 medians. Run 43863642458, tested 2026-11-16; target 4 vCPU · 8 GB · 40 GB. Raw run and harness source
Dimension and measurementStrongest observed result
CPUNode.js web tooling (runs/s)Phoronix Node.js Web Tooling; replicate sandboxesMetric rowMethod23.43 runs/sNamespace5 replicate sandboxes
Memory bandwidthSTREAM Triad (MB/s)STREAM Triad; replicate sandboxesMetric rowMethod260,182 MB/sDaytona (VM)3 replicate sandboxes
Disk I/Ofio rand write 4KB, O_DIRECT (MB/s)fio random write, 4 KB blocks, O_DIRECT; replicate sandboxesMetric rowMethod2,524 MB/sTangle Sandbox3 replicate sandboxes
Network throughputiperf3 WAN download (Mbits/sec)iperf3 WAN download to the nearest curated public server; replicate sandboxesMetric rowMethod8,979 Mbits/secVercel3 replicate sandboxes
Network throughputiperf3 WAN upload (Mbits/sec)iperf3 WAN upload to the nearest curated public server; replicate sandboxesMetric rowMethod18,717 Mbits/secDigitalOcean3 replicate sandboxes
Filesystem behaviorHardlink throughput (bogo ops/s)Hardlink throughput; replicate sandboxesMetric rowMethod18.36 bogo ops/sCreateOS3 replicate sandboxes
Cold workload taskBetter Auth: cold install (s)Better Auth cold dependency install; fresh pipeline sandboxesMetric rowMethod7.85 sDaytona (VM)12 replicate sandboxes
End-to-end workflowBetter Auth: total (s)Better Auth cold pipeline; total sums per-task mediansMetric rowMethod116.4 sModal (VM)12 replicate sandboxes

Precision

Each row shows the single metric with that provider's widest total (min-max) spread, ranked by IDR (p10-p90). The distribution is every trial the run recorded for that metric, pooled across the provider's replicate sandboxes. Phoronix repeats a benchmark several times inside each sandbox, and all of those measurements count, so a row's n is trials rather than sandboxes. Convergence-mode suites repeat until Phoronix settles, which means a noisy sandbox contributes more trials than a quiet one and weighs on the interval accordingly. Metrics that depend on the public Internet, including WAN iperf, repository cloning, and dependency installation, are excluded so the figure reflects sandbox precision; local networking such as iperf loopback remains eligible. The suite and metric vary by provider and are named on every row. Because the units are unlike, every interval is given as a percentage of its own metric's p50. Dots are individual trials and the vertical line marks p50; select Total, IQR, or IDR to highlight and read that range, or hover a row for its numerical endpoints.

Worst precision item per provider

Ranked by IDR (p10-p90) · lower is more precise

Range
Provider · worst itemIDR spread
  1. 1

    Baseten Sandboxes

    Synthetic · STREAM · Copy

    p50 90,016 MB/s · n=6 trials · 3 sandboxes

    IDR spread8.3%
  2. 2

    Novita

    Synthetic · Network · raw.githubusercontent.com HTTPS

    p50 40 ms · n=90 trials · 3 sandboxes

    IDR spread10.8%
  3. 3

    DigitalOcean

    Synthetic · Network · index.crates.io config HTTPS

    p50 10 ms · n=90 trials · 3 sandboxes

    IDR spread18.5%
  4. 4

    Blaxel

    Synthetic · Network · files.pythonhosted.org HTTPS

    p50 12 ms · n=90 trials · 3 sandboxes

    IDR spread24.2%
  5. 5

    microsandbox

    Synthetic · Network · files.pythonhosted.org HTTPS

    p50 13 ms · n=90 trials · 3 sandboxes

    IDR spread26.0%
  6. 6

    CreateOS

    Synthetic · Network · registry.npmjs.org HTTPS

    p50 31 ms · n=90 trials · 3 sandboxes

    IDR spread29.5%
  7. 7

    Vercel

    Synthetic · Network · gcr.io/v2 HTTPS

    p50 19 ms · n=90 trials · 3 sandboxes

    IDR spread36.9%
  8. 8

    Namespace

    Synthetic · Network · registry.npmjs.org HTTPS

    p50 15 ms · n=90 trials · 3 sandboxes

    IDR spread48.2%
  9. 9

    Daytona (container)

    Synthetic · Network · github.com HTTPS

    p50 85 ms · n=90 trials · 3 sandboxes

    IDR spread48.7%
  10. 10

    Tangle Sandbox

    Synthetic · PyBench · execution time

    p50 556 ms · n=6 trials · 3 sandboxes

    IDR spread105%
  11. 11

    Daytona (VM)

    Synthetic · Network · gcr.io/v2 HTTPS

    p50 32 ms · n=90 trials · 3 sandboxes

    IDR spread118%
  12. 12

    Modal (gVisor)

    Synthetic · Network · gcr.io/v2 HTTPS

    p50 31 ms · n=90 trials · 3 sandboxes

    IDR spread210%
  13. 13

    boat

    Synthetic · Network · github.com HTTPS

    p50 63 ms · n=90 trials · 3 sandboxes

    IDR spread280%
  14. 14

    E2B

    Synthetic · Network · Node 22 download

    p50 77 Mbits/sec · n=3 trials · 3 sandboxes

    IDR spread1322%
  15. 15

    Modal (VM)

    Synthetic · Network · github.com HTTPS

    p50 18 ms · n=90 trials · 3 sandboxes

    IDR spread2800%
trialp50selected p10-p90total MIN-MAX

Methodology

Every environment ran the same harness against freshly created sandboxes targeting 4 vCPU · 8 GB RAM · 40 GB disk. Synthetic suites ran on 3 replicate sandboxes (5 for CPU), each repeating passes until Phoronix's own convergence criterion was met, so pass counts vary by how noisy the sandbox was; each real repository pipeline (Better Auth, Mastra, and OpenClaw) ran cold on 12 fresh sandboxes, through that repository's own CI tasks. Every figure is a median (p50) pooled across an environment's sandboxes, and every cell keeps the individual trials behind that median, which is what the Precision section reads.

The 10 synthetic suites are described by their own Phoronix test profiles: 8 published on OpenBenchmarking.org, 2 written for this harness and therefore without a profile page to cite.

NameTest typeDescriptionProject site
Flexible IO Testerv2.1.0DiskFIO, the Flexible I/O Tester, is an advanced Linux disk benchmark supporting multiple I/O engines and a wealth of options. FIO was written by Jens Axboe for testing of the Linux I/O subsystem and schedulers.fio.readthedocs.io
Hardlink Throughputharness-localDiskMeasures hardlink() syscall throughput using the stress-ng --link stressor. Mirrors the per-syscall pattern that pnpm install hits when materialising thousands of node_modules entries from a content-addressable store. Reports bogo ops/sec (real time) — one bogo op = one create+unlink hardlink cycle.github.com/ColinIanKing
Streamv1.3.4MemoryThis is a benchmark of Stream, the popular system memory (RAM) benchmark.cs.virginia.edu
iPerfv1.2.0NetworkiPerf is a network bandwidth throughput testing software. This test profile is used for automated testing of an iperf client and requires you have access to an iperf server.software.es.net
iPerf WANharness-localNetworkThis test measures sustained WAN TCP throughput with iperf3 against an RTT-ranked reachable server from a curated public list, chosen per run by RTT probe. It uses eight parallel streams so that high bandwidth-delay-product paths are not understated, and reports receiver-side goodput.software.es.net
Node.js V8 Web Tooling Benchmarkv1.0.1ProcessorRunning the V8 project's Web-Tooling-Benchmark under Node.js. The Web-Tooling-Benchmark stresses JavaScript-related workloads common to web developers like Babel and TypeScript and Babylon. This test profile can test the system's JavaScript performance with Node.js.v8.github.io
Gitv1.1.0SystemThis test measures the time needed to carry out some sample Git operations on an example, static repository that happens to be a copy of the GNOME GTK tool-kit repository.git-scm.com
PostgreSQLv1.15.0SystemThis is a benchmark of PostgreSQL using the integrated pgbench for facilitating the database benchmarks.postgresql.org
PyBenchv1.1.3SystemThis test profile reports the total time of the different average timed test results from PyBench. PyBench reports average test times for different functions such as BuiltinFunctionCalls and NestedForLoops, with this total result providing a rough estimate as to Python's average performance on a given system. This test profile runs PyBench each time for 20 rounds.svn.python.org
SQLite Speedtestv1.0.1SystemThis is a benchmark of SQLite's speedtest1 benchmark program with an increased problem size of 1,000.sqlite.org
Pipeline totals
The p50 is the sum of per-task medians, so it exactly matches the task numbers reported beside it. In the Results explorer, min, p95, and max come from end-to-end totals aligned within the same replicate sandbox and trial; task percentiles are never added together.
A column is a variant, not a vendor
Modal (gVisor) and Modal (VM) are one account's two runtimes booting the same image. Variants attempted without validated metrics are disclosed under Data quality rather than rendered as empty columns.
A column is a fleet, not a machine
Each cell pools an environment's replicate sandboxes, and the environments table flags every provider whose sandboxes reported more than one host CPU model.
n counts trials, r counts sandboxes
A convergence-mode cell's n is pooled passes, not sandboxes. A sandbox that converged slowly contributes more passes to the same median. Where a sandbox or a whole suite was lost, the surviving replicates still carry the cell, which is why the replicate count in the summary above is a range (1 to 12) rather than one number; Data quality records every such loss.
Network figures isolate the sandbox
iperf3 loopback, using TCP at 1 and 10 streams plus a 10 G-objective UDP datagram test, measures the sandbox's own network stack with no Internet path. The WAN pair measures both directions against the nearest curated public iperf3 server, chosen per run by RTT probe.
Task names are each repository's own
Labels come from the dataset's catalog, which uses the repo's names: Better Auth's lint types (its repo-wide type-lint pass) is a different task from its root typecheck.

All metrics

Every catalogued metric this run emitted, labeled and grouped exactly as the dataset's own leaderboard labels it. Each cell shows the median and its ratio to the row's best; amber deepens the further a provider sits behind, teal marks the best. ↓ means lower is better; ∑ marks a value derived from other rows. Click a metric name to open it in the results explorer.

MetricBaseten SandboxesBlaxelboatCreateOSDaytona (container)Daytona (VM) †DigitalOcean †E2BmicrosandboxModal (gVisor) †Modal (VM) †NamespaceNovitaTangle SandboxVercelSpread
CPU
runs/s ↑headline20.67×1.1320.14×1.1618.02×1.3021.43×1.0916.26×1.4421.73×1.0821.27×1.1014.29×1.6416.77×1.4020.11×1.1721.52×1.0923.43best19.14×1.228.34×2.8111.25×2.08×2.8
Disk I/O
IOPS ↑–250,000×2.56296,500×2.1652,400×12.22110,500×5.80170,500×3.76117,000×5.4759,850×10.70178,500×3.59171,000×3.75156,500×4.0996,900×6.6179,150×8.09640,500best269,000×2.38×12.2
MB/s ↑–1,022×2.571,214×2.16215×12.21452×5.80698×3.76478×5.49245×10.72729×3.60700×3.75642×4.09397×6.61324×8.102,625best1,103×2.38×12.2
IOPS ↑–201,000×3.06194,000×3.1885,100×7.2456,450×10.91217,500×2.83103,700×5.9459,200×10.4160,600×10.17189,000×3.26164,000×3.76112,500×5.4878,650×7.83616,000best305,000×2.02×10.9
MB/s ↑headline–823×3.07794×3.18349×7.24231×10.92892×2.83425×5.94242×10.42248×10.18774×3.26673×3.75460×5.48322×7.842,524best1,250×2.02×10.9
IOPS ↑–5,044×20.826,804×15.432,804×37.459,636×10.9021,000×5.002,835×37.04600×175.158,097×12.97105,000best13,850×7.582,315×45.378,746×12.016,887×15.253,618×29.02×175.1
MB/s ↑–5,291×20.807,136×15.422,941×37.4210,106×10.8922,065×4.992,973×37.02630×174.648,492×12.96110,059best14,549×7.562,429×45.329,172×12.007,223×15.243,795×29.00×174.6
IOPS ↑–5,897×2.791,719×9.572,604×6.324,031×4.0816,450best2,965×5.55599×27.466,293×2.619,506×1.739,213×1.791,899×8.665,696×2.895,683×2.893,303×4.98×27.5
MB/s ↑–6,186×2.791,803×9.592,732×6.334,228×4.0917,287best3,111×5.56630×27.436,601×2.629,971×1.739,685×1.791,992×8.685,974×2.895,960×2.903,464×4.99×27.4
Memory bandwidth
MB/s ↑67,053×3.8467,149×3.8456,923×4.5344,871×5.7418,320×14.07257,770best140,930×1.8348,943×5.2756,521×4.56145,003×1.78144,573×1.7833,145×7.78156,337×1.6524,382×10.5751,896×4.97×14.1
MB/s ↑90,016×3.0890,024×3.0875,319×3.6847,805×5.7922,769×12.16276,947best170,607×1.6274,259×3.7386,565×3.20191,185×1.45190,610×1.4544,070×6.28162,339×1.7130,439×9.1083,480×3.32×12.2
MB/s ↑61,828×3.9161,709×3.9249,594×4.8841,219×5.8722,992×10.52241,802best135,359×1.7943,568×5.5552,197×4.63134,631×1.80134,404×1.8030,162×8.02150,230×1.6119,912×12.1444,430×5.44×12.1
MB/s ↑headline67,417×3.8667,519×3.8557,946×4.4945,010×5.7822,990×11.32260,182best141,986×1.8348,927×5.3256,590×4.60146,181×1.78150,282×1.7333,233×7.83155,231×1.6824,104×10.7952,327×4.97×11.3
Network
Mbits/sec ↑–69,027×2.2163,225×2.4185,644×1.7833,911×4.4983,811×1.82113,332×1.3461,603×2.4759,413×2.5640,891×3.7258,464×2.6034,751×4.38152,238best–64,076×2.38×4.5
Mbits/sec ↑–89,540×1.8655,152×3.02127,736×1.3137,217×4.4898,988×1.68166,699best48,830×3.4177,077×2.1635,913×4.6473,000×2.2838,171×4.37150,864×1.10–47,567×3.50×4.6
Mbits/sec ↑–9,999best9,999best9,999best9,999best9,999best9,999best9,999best9,999best517×19.349,999best9,999best9,999best–9,999best×19.3
Mbits/sec ↑headline–3,569×2.524,775×1.88266×33.813,868×2.326,386×1.414,292×2.091,595×5.631,052×8.542,225×4.031,676×5.367,047×1.275,066×1.77–8,979best×33.8
Mbits/sec ↑headline–2,429×7.701,831×10.22920×20.353,767×4.974,878×3.8418,717best871×21.491,055×17.741,597×11.724,356×4.304,222×4.434,203×4.45–5,550×3.37×21.5
ms ↓–27×1.5163×3.5642×2.3585×4.8019×1.06313×17.62106×5.98208×11.71516×29.0418best20×1.12518×29.18–26×1.48×29.2
ms ↓–15×1.3022×1.9123×2.0439×3.4412×1.09122×10.7946×4.07102×9.0326×2.3411best12×1.03194×17.21–16×1.43×17.2
ms ↓–12×1.3923×2.6421×2.4018×2.1110×1.1710×1.1932×3.6614×1.6416×1.869best9×1.0640×4.67–14×1.61×4.7
ms ↓–21×1.3837×2.4631×2.0528×1.8722×1.4617×1.1247×3.1457×3.8024×1.6019×1.2715best62×4.09–23×1.50×4.1
ms ↓–14×1.4129×2.8922×2.2720×2.0511×1.1212×1.1933×3.3116×1.5818×1.8310×1.0310best45×4.52–15×1.56×4.5
ms ↓–12×1.4123×2.7421×2.4618×2.2010×1.2310×1.2133×3.9013×1.5916×1.959×1.028best40×4.80–14×1.64×4.8
ms ↓–21×1.12128×6.8026×1.4195×5.0420×1.05131×6.95112×5.96120×6.4090×4.8119best20×1.06203×10.77–21×1.10×10.8
ms ↓–17×1.16530×36.5361×4.2241×2.8432×2.19118×8.1420×1.3688×6.0431×2.1127×1.8915best19×1.28–19×1.32×36.5
ms ↓–32×1.13143×4.9742×1.48109×3.7833×1.1564×2.24137×4.7754×1.86110×3.8432×1.1229best126×4.40–36×1.26×5.0
ms ↓–17×1.2434×2.4826×1.9123×1.7015×1.0914×1.0134×2.4416×1.1420×1.4414best14×1.0143×3.14–19×1.39×3.1
ms ↓–12×1.3624×2.7421×2.4021×2.4611×1.2510×1.1831×3.6413×1.4616×1.839×1.019best40×4.59–14×1.67×4.6
ms ↓–0best5×Infinity8×Infinity6×Infinity5×Infinity16×Infinity12×Infinity4×Infinity17×Infinity4×Infinity4×Infinity16×Infinity–3×Infinity×Infinity
ms ↓–4×1.3316×5.3312×4.007×2.3312×4.008×2.6712×4.008×2.6715×5.004×1.334×1.3316×5.33–3best×5.3
ms ↓–4×Infinity18×Infinity8×Infinity6×Infinity5×Infinity44×Infinity24×Infinity4×Infinity58×Infinity4×Infinity4×Infinity52×Infinity–0best×Infinity
ms ↓–4×1.335×1.6720×6.6714×4.6718×6.0060×20.0032×10.6780×26.6792×30.678×2.6736×12.00120×40.00–3best×40.0
ms ↓–55×3.4416best59×3.6959×3.6959×3.6936×2.2580×5.00160×10.0098×6.1360×3.7556×3.5088×5.50–51×3.19×10.0
Mbits/sec ↑–2,885×1.021,255×2.35871×3.381,430×2.062,675×1.102,852×1.0377×38.202,074×1.421,655×1.781,032×2.852,744×1.071,066×2.76–2,945best×38.2
System
s ↓headline–51.1×1.3946.9×1.2837.2×1.0146.4×1.2760.6×1.6540.8×1.1169.2×1.8944.0×1.2042.4×1.1638.6×1.0536.7best44.2×1.2154.4×1.4863.6×1.74×1.9
TPS ↑–256,218×4.29216,992×5.07285,891×3.8596,850×11.35670,871×1.64617,554×1.78156,149×7.04218,046×5.04638,970×1.721,099,303best238,631×4.61282,859×3.89–177,008×6.21×11.4
ms ↓–0×4.300×5.070×3.851×11.340×1.650×1.770×7.050×5.030×1.710best0×4.600×3.89–0×6.21×11.3
TPS ↑–20,550×3.2311,050×6.0216,216×4.106,376×10.4338,878×1.7120,128×3.309,465×7.0216,274×4.0844,744×1.4966,476best20,543×3.2425,177×2.64–16,271×4.09×10.4
ms ↓–2×3.245×6.033×4.108×10.431×1.712×3.305×7.033×4.091×1.491best2×3.242×2.65–3×4.09×10.4
ms ↓–528×1.44431×1.18553×1.51599×1.64432×1.18444×1.21804×2.20475×1.30439×1.20436×1.19366best482×1.32556×1.52766×2.09×2.2
s ↓–48.3×1.4681.6×2.4735.4×1.07105.9×3.2035.6×1.0839.7×1.2075.7×2.2954.4×1.6585.4×2.5851.1×1.5433.1best42.0×1.2756.5×1.7168.4×2.07×3.2
Real-world pipelines
s ↓ ∑–198.8×1.71187.1×1.61209.4×1.80–121.4×1.04149.0×1.28305.4×2.62228.4×1.96157.4×1.35116.4best178.2×1.53192.3×1.65–277.6×2.38×2.6
s ↓–0.70×1.271.72×3.111.78×3.22–0.62×1.121.42×2.563.23×5.851.11×2.020.80×1.460.62×1.130.55best2.08×3.77–0.85×1.54×5.8
s ↓–17.1×2.1715.8×2.0128.1×3.58–7.85best15.4×1.9621.6×2.7614.9×1.9019.5×2.4912.7×1.6210.7×1.3613.3×1.69–19.2×2.44×3.6
s ↓–3.57×2.693.15×2.373.24×2.43–1.79×1.342.04×1.535.06×3.813.90×2.935.96×4.481.33best2.82×2.123.34×2.51–4.42×3.32×4.5
s ↓–11.4×1.1810.8×1.1111.9×1.23–9.69best11.6×1.1919.2×1.9813.7×1.4112.8×1.339.91×1.0211.6×1.2011.5×1.19–16.5×1.70×2.0
s ↓–3.05×1.192.78×1.092.94×1.15–2.56best2.72×1.065.21×2.043.63×1.422.92×1.142.66×1.042.77×1.083.06×1.20–4.95×1.94×2.0
s ↓–8.97×1.397.06×1.097.23×1.12–6.46best6.97×1.0813.6×2.109.71×1.507.70×1.196.49×1.009.30×1.447.78×1.20–12.6×1.95×2.1
s ↓–33.9×3.4331.9×3.2238.2×3.86–14.0×1.4116.0×1.6152.1×5.2641.9×4.2322.0×2.229.90best35.8×3.6233.8×3.41–52.4×5.29×5.3
s ↓–3.04×1.942.84×1.812.56×1.63–1.71×1.091.67×1.064.28×2.733.45×2.202.14×1.371.57best2.33×1.482.74×1.74–4.03×2.57×2.7
s ↓–48.2×1.3645.0×1.2745.3×1.28–36.2×1.0237.4×1.0677.6×2.1954.0×1.5340.7×1.1535.4best43.5×1.2346.2×1.31–67.7×1.92×2.2
s ↓–68.9×1.9266.2×1.8568.0×1.90–40.6×1.1353.8×1.50103.5×2.8982.0×2.2942.9×1.2035.9best58.8×1.6468.5×1.91–94.9×2.65×2.9
s ↓ ∑–1302.0×1.201356.5×1.251126.6×1.04–1084.7best1164.0×1.071804.4×1.661289.3×1.191269.7×1.171111.5×1.021134.4×1.051251.7×1.15–1646.6×1.52×1.7
s ↓–1.93×1.284.98×3.314.81×3.20–1.75×1.164.37×2.903.45×2.292.30×1.532.14×1.421.68×1.121.50best3.23×2.14–2.42×1.61×3.3
s ↓headline–44.9×1.3762.4×1.9161.5×1.88–32.6best48.2×1.4867.1×2.0655.5×1.7044.7×1.3739.3×1.2037.5×1.1547.7×1.46–61.2×1.87×2.1
s ↓–101.7×1.21101.7×1.2194.5×1.12–84.3best87.2×1.03156.9×1.86121.9×1.4596.5×1.1588.3×1.0596.6×1.15100.4×1.19–160.1×1.90×1.9
s ↓–83.0×1.4093.4×1.5775.2×1.27–60.8×1.0370.5×1.19131.3×2.2187.8×1.4868.4×1.1559.3best65.3×1.1080.0×1.35–120.2×2.03×2.2
s ↓–1070.6×1.201094.0×1.23890.6best–905.2×1.02953.7×1.071445.6×1.621021.8×1.151058.0×1.19922.9×1.04933.4×1.051020.4×1.15–1302.8×1.46×1.6
s ↓ ∑–677.7×1.69646.7×1.62571.0×1.43–400.2best802.3×2.001169.3×2.92615.1×1.54545.8×1.36557.0×1.39669.4×1.67746.2×1.86–825.3×2.06×2.9
s ↓–3.03×1.326.23×2.724.72×2.06–2.77×1.218.98×3.916.04×2.633.76×1.643.67×1.602.76×1.202.29best5.29×2.31–3.80×1.66×3.9
s ↓–15.8×2.0617.9×2.3420.2×2.65–7.64best29.1×3.8123.2×3.0417.4×2.2814.4×1.8911.2×1.4711.6×1.5215.1×1.98–18.6×2.43×3.8
s ↓–340.0×1.90318.2×1.77285.8×1.59–179.4best370.5×2.07599.5×3.34308.6×1.72231.5×1.29239.1×1.33350.4×1.95389.6×2.17–410.6×2.29×3.3
s ↓–177.1×1.69165.6×1.58139.6×1.33–104.6best209.9×2.01313.8×3.00156.0×1.49138.2×1.32133.5×1.28167.8×1.60192.8×1.84–218.5×2.09×3.0
s ↓–20.6×1.8919.2×1.7719.3×1.78–10.9best22.1×2.0336.6×3.3618.3×1.6812.1×1.1112.6×1.1621.0×1.9322.9×2.11–26.8×2.47×3.4
s ↓–121.3×1.28119.5×1.26101.4×1.07–94.9best161.8×1.70190.2×2.00110.9×1.17146.0×1.54157.9×1.66116.4×1.23120.4×1.27–147.0×1.55×2.0

Observed environments

Specs, storage shape, hardware mixtures, and egress identity come directly from the run's embedded in-sandbox probes and Phoronix metadata. When replicate sandboxes reported different host CPU models, the CPU row shows the fleet count and the disclosure below lists every model. Environment flags mark an off-target resource, gVisor isolation, or a heterogeneous fleet; missing probe values remain blank rather than being filled from another run.

SpecBaseten SandboxesBlaxelboatCreateOSDaytona (container)Daytona (VM) †DigitalOcean †E2BmicrosandboxModal (gVisor) †Modal (VM) †NamespaceNovitaTangle SandboxVercel
vCPUs44444884420204444
CPU modelIntel(R) Xeon(R) ProcessorIntel(R) Xeon(R) Processor4 models across replicates ⚠AMD EPYC2 models across replicates ⚠AMD EPYCAMD EPYCIntel(R) Xeon(R) Processor @ 2.60GHzAMD EPYCunknown3 models across replicates ⚠AMD EPYCAMD EPYC4 models across replicates ⚠2 models across replicates ⚠
Isolationfirecrackerfirecrackerqemu-kvmfirecrackerdocker-runcfirecrackercontainerd-runcfirecrackerlibkrungvisorcloud-hypervisoroci-containerfirecrackerdocker-runcfirecracker
Virtualization–firecrackerqemu-kvmfirecrackerunknownfirecrackerfirecrackerfirecrackerlibkrunnot-observablecloud-hypervisorfirecrackerfirecrackerunknownfirecracker
Memory (GB)7.787.787.767.7887.7715.647.777.7887.767.817.7888
Disk (GB)40.240.278.439.310250.9310.4104.239.351250462.9103.6466.363.9
File systemfusexfsext4overlayfsoverlayfsoverlayfsoverlayfsext4ext4overlayfsext4overlayfsext4tmpfsxfs
Mount options––relatime,rw––––discard,relatime,rwrelatime,rw–discard,noatime,prjquota,rw–relatime,rw–inode64,logbsize=32k,logbufs=8,noatime,noquota,nouuid,rw
I/O scheduler––NONE–NONE–––NONE––––NONEMQ-DEADLINE
Block size––4096––––40964096–4096–4096–4096
Kernel6.12.1146.12.1146.8.0-117-generic7.3.16.8.0-90-generic6.18.576.1.1766.1.158+6.12.1134.19.0-gvisor7.3.47.1.46.1.158+6.8.0-147-generic6.18.56
OSDebian GNU/Linux 13 (trixie)Debian GNU/Linux 13 (trixie)Ubuntu 24.04.4 LTSUbuntu 24.04.4 LTSDebian GNU/Linux 13 (trixie)Debian GNU/Linux 13 (trixie)Debian GNU/Linux 13 (trixie)Debian GNU/Linux 13 (trixie)Debian GNU/Linux 13 (trixie)Debian GNU/Linux 13 (trixie)Debian GNU/Linux 13 (trixie)Debian GNU/Linux 13 (trixie)Debian GNU/Linux 13 (trixie)Debian GNU/Linux 13 (trixie)Debian GNU/Linux 13 (trixie)
Egress family–IPv6IPv4IPv4IPv4IPv4IPv4IPv4IPv4IPv4IPv4IPv4IPv4IPv4IPv4
Egress ASN–AS14618AS16276AS16276AS396356AS398465AS14061AS396982AS20473AS400494AS398465AS401483AS31898AS16276AS16509
ASN name–Amazon.com, Inc.OVH SASOVH SASLatitude.shRackdog, LLCDigitalOcean, LLCGoogle LLCThe Constant Company, LLCCirrascale Cloud Services LLCRackdog, LLCNamespace LabsOracle CorporationOVH SASAmazon.com, Inc.
Geo location–Ashburn, Virginia, USRoubaix, Hauts-de-France, FRWarrenton, Virginia, USNew York City, New York, USAshburn, Virginia, USKansas City, Missouri, USThe Dalles, Oregon, USElk Grove Village, Illinois, USSeattle, Washington, USNew York City, New York, USAshburn, Virginia, USPhoenix, Arizona, USWarrenton, Virginia, USAshburn, Virginia, US
Spec vs targetmatchedmatchedmatchedmatchedmatchedmismatchmismatchmatchedmatchedmismatchmismatchmatchedmatchedmatchedmatched

⚠ Heterogeneous fleets: boat saw 4 host CPU models across its replicate sandboxes (AMD EPYC 4585PX 16-Core Processor; AMD EPYC 9455 48-Core Processor; AMD Ryzen 7 9700X 8-Core Processor; AMD Ryzen 9 9950X 16-Core Processor) · Daytona (container) saw 2 host CPU models across its replicate sandboxes (AMD EPYC 9254 24-Core Processor; AMD EPYC 9354P 32-Core Processor) · Modal (VM) saw 3 host CPU models across its replicate sandboxes (AMD EPYC 9655 96-Core Emb Processor; AMD EPYC 9655 96-Core Processor; AMD EPYC 9655P 96-Core Processor) · Tangle Sandbox saw 4 host CPU models across its replicate sandboxes (Intel(R) Core(TM) i7-8700 CPU @ 3.20GHz; Intel(R) Core(TM) i9-9900K CPU @ 3.60GHz; Intel(R) Xeon(R) CPU E5-2687W v4 @ 3.00GHz; Intel(R) Xeon(R) E-2386G CPU @ 3.50GHz) · Vercel saw 2 host CPU models across its replicate sandboxes (Intel(R) Xeon(R) Processor @ 2.50GHz; Intel(R) Xeon(R) Processor @ 2.90GHz). Every median for each of these environments pools across those machines, so read their cells as the fleet's typical draw, not one machine's speed.

Data quality

Only providers marked validated by the dataset render as comparison columns. Every skipped or failed suite record below comes from this run, including universally unrecorded tasks and provider-specific failures; no older run fills a missing cell. Providers attempted without any validated metrics are disclosed after the table instead of becoming empty comparison columns.

ProviderSuiteOutcomeReason
Daytona (container)realworld-better-authskipped · diskInsufficient disk: 10.0 GiB free, suite needs 10 GiB
Daytona (container)realworld-mastraskipped · diskInsufficient disk: 10.0 GiB free, suite needs 30 GiB
Daytona (container)realworld-openclawskipped · diskInsufficient disk: 10.0 GiB free, suite needs 25 GiB
Tangle Sandboxpgbenchskippedpts_pgbench-read-only: could not claim the root-owned baked postgres data dir (needs sudo)
Tangle Sandboxpgbenchskippedpts_pgbench-read-write: could not claim the root-owned baked postgres data dir (needs sudo)
Tangle Sandboxrealworld-better-authskippedpts_realworld-better-auth: PTS install of local/realworld-better-auth-1.0.0 failed (exit 0, not in list-installed-tests)
Tangle Sandboxrealworld-mastraskippedpts_realworld-mastra: PTS install of local/realworld-mastra-1.0.0 failed (exit 0, not in list-installed-tests)
Tangle Sandboxrealworld-openclawskippedpts_realworld-openclaw: PTS install of local/realworld-openclaw-1.0.0 failed (exit 0, not in list-installed-tests)
Baseten Sandboxescpu-nodefailedFailed to create sandbox: computesdk create failed: sandbox API error (HTTP 403) FORBIDDEN: This API key does not have permission to perform this operation.
Baseten Sandboxescpu-nodefailedPartial publication withheld unverified measurements: node_web_tooling_runs_per_s
Baseten SandboxesdiskfailedFailed to create sandbox: computesdk create failed: sandbox API error (HTTP 403) FORBIDDEN: This API key does not have permission to perform this operation.
Baseten SandboxesdiskfailedPartial publication withheld unverified measurements: hardlink_bogo_ops_per_s, fio_type_sequential_read_engine_linux_aio_direct_yes_block_size_1mb_job_count_1_disk_target_default_test_directory_mb_per_s, fio_type_sequential_read_engine_linux_aio_direct_yes_block_size_1mb_job_count_1_disk_target_default_test_directory_iops, fio_type_sequential_write_engine_linux_aio_direct_yes_block_size_1mb_job_count_1_disk_target_default_test_directory_mb_per_s, fio_type_sequential_write_engine_linux_aio_direct_yes_block_size_1mb_job_count_1_disk_target_default_test_directory_iops, fio_type_random_read_engine_linux_aio_direct_yes_block_size_4kb_job_count_1_disk_target_default_test_directory_mb_per_s, fio_type_random_read_engine_linux_aio_direct_yes_block_size_4kb_job_count_1_disk_target_default_test_directory_iops, fio_type_random_write_engine_linux_aio_direct_yes_block_size_4kb_job_count_1_disk_target_default_test_directory_mb_per_s, fio_type_random_write_engine_linux_aio_direct_yes_block_size_4kb_job_count_1_disk_target_default_test_directory_iops
Baseten SandboxesnetworkfailedFailed to create sandbox: computesdk create failed: sandbox API error (HTTP 403) FORBIDDEN: This API key does not have permission to perform this operation.
Baseten SandboxesnetworkfailedPartial publication withheld unverified measurements: iperf_server_address_localhost_server_port_5201_duration_10_seconds_test_tcp_parallel_1, iperf_server_address_localhost_server_port_5201_duration_10_seconds_test_tcp_parallel_10, iperf_server_address_localhost_server_port_5201_duration_10_seconds_test_udp_10000mbit_objective_parallel_1, iperf_wan_direction_download, iperf_wan_direction_upload, network_https_github_com_total_ms, network_https_api_github_com_total_ms, network_https_raw_githubusercontent_com_total_ms, network_https_registry_npmjs_org_total_ms, network_https_pypi_org_total_ms, network_https_files_pythonhosted_org_total_ms, network_https_ghcr_io_v2_total_ms, network_https_gcr_io_v2_total_ms, network_https_auth_docker_io_registry_token_total_ms, network_https_crates_io_total_ms, network_https_index_crates_io_config_total_ms, network_dns_cold_github_com_ms, network_dns_cold_registry_npmjs_org_ms, network_dns_cold_docker_io_ms, network_dns_cold_pypi_org_ms, network_dns_cold_rubygems_org_ms, network_download_node_v22_23_1_linux_x64_mbits_per_sec
Baseten SandboxespgbenchfailedFailed to create sandbox: computesdk create failed: sandbox API error (HTTP 403) FORBIDDEN: This API key does not have permission to perform this operation.
Baseten SandboxespgbenchfailedPartial publication withheld unverified measurements: pgbench_scaling_factor_100_clients_50_mode_read_only, pgbench_scaling_factor_100_clients_50_mode_read_only_average_latency, pgbench_scaling_factor_100_clients_50_mode_read_write, pgbench_scaling_factor_100_clients_50_mode_read_write_average_latency
Baseten Sandboxesrealworld-better-authfailedFailed to create sandbox: computesdk create failed: sandbox API error (HTTP 403) FORBIDDEN: This API key does not have permission to perform this operation.
Baseten Sandboxesrealworld-better-authfailedPartial publication withheld unverified measurements: realworld_better_auth_task_git_clone, realworld_better_auth_task_cold_install, realworld_better_auth_task_lint_biome, realworld_better_auth_task_lint_deps_knip, realworld_better_auth_task_lint_format, realworld_better_auth_task_lint_spell, realworld_better_auth_task_lint_types, realworld_better_auth_task_lint_packages, realworld_better_auth_task_typecheck, realworld_better_auth_task_build
Baseten Sandboxesrealworld-better-authfailedStep "mise run benchmark:realworld:pts:better-auth" lost its sandbox: 12 consecutive detached polls failed (last: computesdk filesystem exists failed: sandbox API error (HTTP 403)) — the sandbox stopped responding, not a quiet long step
Baseten Sandboxesrealworld-mastrafailedStep "REALWORLD_TASK_TIMEOUT_SECONDS=2400 mise run benchmark:realworld:pts:mastra" lost its sandbox: 12 consecutive detached polls failed (last: computesdk filesystem exists failed: sandbox API error (HTTP 403)) — the sandbox stopped responding, not a quiet long step
Baseten Sandboxesrealworld-mastrafailedPartial publication withheld unverified measurements: realworld_mastra_task_git_clone, realworld_mastra_task_cold_install, realworld_mastra_task_lint_format, realworld_mastra_task_build_core, realworld_mastra_task_test_core
Baseten Sandboxesrealworld-mastrafailedFailed to create sandbox: computesdk create failed: sandbox API error (HTTP 403) FORBIDDEN: This API key does not have permission to perform this operation.
Baseten Sandboxesrealworld-openclawfailedFailed to create sandbox: computesdk create failed: sandbox API error (HTTP 403) FORBIDDEN: This API key does not have permission to perform this operation.
Baseten Sandboxesrealworld-openclawfailedPartial publication withheld unverified measurements: realworld_openclaw_task_git_clone, realworld_openclaw_task_cold_install, realworld_openclaw_task_lint_oxlint, realworld_openclaw_task_lint_extensions_all, realworld_openclaw_task_typecheck, realworld_openclaw_task_test_types
Baseten SandboxessystemfailedFailed to create sandbox: computesdk create failed: sandbox API error (HTTP 403) FORBIDDEN: This API key does not have permission to perform this operation.
Baseten SandboxessystemfailedPartial publication withheld unverified measurements: pybench_milliseconds, sqlite_speedtest_seconds, git_seconds
boatrealworld-mastrafailedPTS ran but every trial failed for 1 of 5 declared metrics: realworld_mastra_task_test_core (realworld-mastra/pts_realworld-mastra.xml) — attempted, no value recorded
boatrealworld-mastrafailedPartial publication withheld unverified measurements: realworld_mastra_task_test_core
CreateOSrealworld-openclawfailedFailed to create sandbox: computesdk create failed: service unavailable
CreateOSrealworld-openclawfailedPartial publication withheld unverified measurements: realworld_openclaw_task_git_clone, realworld_openclaw_task_cold_install, realworld_openclaw_task_lint_oxlint, realworld_openclaw_task_lint_extensions_all, realworld_openclaw_task_typecheck, realworld_openclaw_task_test_types
Daytona (container)realworld-better-authfailedPartial publication withheld unverified measurements: realworld_better_auth_task_git_clone, realworld_better_auth_task_cold_install, realworld_better_auth_task_lint_biome, realworld_better_auth_task_lint_deps_knip, realworld_better_auth_task_lint_format, realworld_better_auth_task_lint_spell, realworld_better_auth_task_lint_types, realworld_better_auth_task_lint_packages, realworld_better_auth_task_typecheck, realworld_better_auth_task_build
Daytona (container)realworld-mastrafailedPartial publication withheld unverified measurements: realworld_mastra_task_git_clone, realworld_mastra_task_cold_install, realworld_mastra_task_lint_format, realworld_mastra_task_build_core, realworld_mastra_task_test_core
Daytona (container)realworld-openclawfailedPartial publication withheld unverified measurements: realworld_openclaw_task_git_clone, realworld_openclaw_task_cold_install, realworld_openclaw_task_lint_oxlint, realworld_openclaw_task_lint_extensions_all, realworld_openclaw_task_typecheck, realworld_openclaw_task_test_types
microsandboxrealworld-openclawfailedFailed to create sandbox: computesdk create failed: runtime error: cloud sandbox "bench-cloud-198c006d-63eb-425c-b53e-2dc641c6b43c" did not reach running before the readiness wait expired
microsandboxrealworld-openclawfailedPartial publication withheld unverified measurements: realworld_openclaw_task_git_clone, realworld_openclaw_task_cold_install, realworld_openclaw_task_lint_oxlint, realworld_openclaw_task_lint_extensions_all, realworld_openclaw_task_typecheck, realworld_openclaw_task_test_types
microsandboxrealworld-openclawfailedFailed to create sandbox: computesdk create failed: runtime error: cloud sandbox "bench-cloud-b1f2648a-488f-446e-9249-0e5803bac324" did not reach running before the readiness wait expired
microsandboxrealworld-openclawfailedFailed to create sandbox: computesdk create failed: runtime error: cloud sandbox "bench-cloud-3027dc27-ae1a-4265-a2b8-5ef09bbad38e" did not reach running before the readiness wait expired
microsandboxrealworld-openclawfailedFailed to create sandbox: computesdk create failed: runtime error: cloud sandbox "bench-cloud-adecad41-020c-47b4-9c97-f997ab29cd71" did not reach running before the readiness wait expired
microsandboxrealworld-openclawfailedFailed to create sandbox: computesdk create failed: runtime error: cloud sandbox "bench-cloud-53e936ef-717a-45ab-a099-ccccf9992ce4" did not reach running before the readiness wait expired
microsandboxrealworld-openclawfailedFailed to create sandbox: computesdk create failed: runtime error: cloud sandbox "bench-cloud-dd2564dd-150b-4029-b071-87f951f3b156" did not reach running before the readiness wait expired
microsandboxrealworld-openclawfailedFailed to create sandbox: computesdk create failed: runtime error: cloud sandbox "bench-cloud-b48b9def-f970-462f-ab54-10c6df8d7ee2" did not reach running before the readiness wait expired
microsandboxrealworld-openclawfailedFailed to create sandbox: computesdk create failed: runtime error: cloud sandbox "bench-cloud-54b853c3-80d5-4a0b-851f-f753324e3cde" did not reach running before the readiness wait expired
microsandboxrealworld-openclawfailedFailed to create sandbox: computesdk create failed: runtime error: cloud sandbox "bench-cloud-4d4b401c-403b-469a-bc7f-01c0bcbb2d1a" did not reach running before the readiness wait expired
Modal (gVisor) †realworld-mastrafailedPTS ran but every trial failed for 1 of 5 declared metrics: realworld_mastra_task_test_core (realworld-mastra/pts_realworld-mastra.xml) — attempted, no value recorded
Modal (gVisor) †realworld-mastrafailedPartial publication withheld unverified measurements: realworld_mastra_task_test_core
Modal (VM) †realworld-mastrafailedFailed to create sandbox: computesdk created-request preparation and verification failed: Modal control operation exceeded 5000ms
Modal (VM) †realworld-mastrafailedPartial publication withheld unverified measurements: realworld_mastra_task_git_clone, realworld_mastra_task_cold_install, realworld_mastra_task_lint_format, realworld_mastra_task_build_core, realworld_mastra_task_test_core
Tangle Sandboxnetworkfailedpts_iperf-wan-download: pts_iperf-wan-download did not produce 1 numeric metric value(s)
Tangle Sandboxnetworkfailedpts_iperf-wan-upload: PTS batch-run of local/iperf-wan-1.0.0 completed but every trial errored (composite carries no values)
Tangle SandboxnetworkfailedStep "mise run benchmark:network:suite" failed with exit code 1
Tangle SandboxnetworkfailedPartial publication withheld unverified measurements: iperf_server_address_localhost_server_port_5201_duration_10_seconds_test_tcp_parallel_1, iperf_server_address_localhost_server_port_5201_duration_10_seconds_test_tcp_parallel_10, iperf_server_address_localhost_server_port_5201_duration_10_seconds_test_udp_10000mbit_objective_parallel_1, iperf_wan_direction_download, iperf_wan_direction_upload, network_https_github_com_total_ms, network_https_api_github_com_total_ms, network_https_raw_githubusercontent_com_total_ms, network_https_registry_npmjs_org_total_ms, network_https_pypi_org_total_ms, network_https_files_pythonhosted_org_total_ms, network_https_ghcr_io_v2_total_ms, network_https_gcr_io_v2_total_ms, network_https_auth_docker_io_registry_token_total_ms, network_https_crates_io_total_ms, network_https_index_crates_io_config_total_ms, network_dns_cold_github_com_ms, network_dns_cold_registry_npmjs_org_ms, network_dns_cold_docker_io_ms, network_dns_cold_pypi_org_ms, network_dns_cold_rubygems_org_ms, network_download_node_v22_23_1_linux_x64_mbits_per_sec
Tangle SandboxpgbenchfailedSuite "pgbench" on tangle produced no pts_*.xml — PTS likely failed silently
Tangle SandboxpgbenchfailedPartial publication withheld unverified measurements: pgbench_scaling_factor_100_clients_50_mode_read_only, pgbench_scaling_factor_100_clients_50_mode_read_only_average_latency, pgbench_scaling_factor_100_clients_50_mode_read_write, pgbench_scaling_factor_100_clients_50_mode_read_write_average_latency
Tangle Sandboxrealworld-better-authfailedSuite "realworld-better-auth" on tangle produced no pts_*.xml — PTS likely failed silently
Tangle Sandboxrealworld-better-authfailedPartial publication withheld unverified measurements: realworld_better_auth_task_git_clone, realworld_better_auth_task_cold_install, realworld_better_auth_task_lint_biome, realworld_better_auth_task_lint_deps_knip, realworld_better_auth_task_lint_format, realworld_better_auth_task_lint_spell, realworld_better_auth_task_lint_types, realworld_better_auth_task_lint_packages, realworld_better_auth_task_typecheck, realworld_better_auth_task_build
Tangle Sandboxrealworld-mastrafailedSuite "realworld-mastra" on tangle produced no pts_*.xml — PTS likely failed silently
Tangle Sandboxrealworld-mastrafailedPartial publication withheld unverified measurements: realworld_mastra_task_git_clone, realworld_mastra_task_cold_install, realworld_mastra_task_lint_format, realworld_mastra_task_build_core, realworld_mastra_task_test_core
Tangle Sandboxrealworld-openclawfailedSuite "realworld-openclaw" on tangle produced no pts_*.xml — PTS likely failed silently
Tangle Sandboxrealworld-openclawfailedPartial publication withheld unverified measurements: realworld_openclaw_task_git_clone, realworld_openclaw_task_cold_install, realworld_openclaw_task_lint_oxlint, realworld_openclaw_task_lint_extensions_all, realworld_openclaw_task_typecheck, realworld_openclaw_task_test_types

Also in this run, not in the tables above: Brezel: pending, 0 metrics reported; Freestyle: pending, 0 metrics reported; run.cloud: pending, 0 metrics reported; tama: pending, 0 metrics reported. An environment that reported nothing gets no column.