Update benchmark behavior

[HeyGarrison]

Summary

• Minor benchmark reliability updates.

[github-actions]

Sandbox Benchmark Results

Sequential

#	Provider	Score	Median TTI	P95	P99	Status
1	declaw	98.9	0.05s	0.21s	0.21s	10/10
2	daytona	97.8	0.19s	0.27s	0.27s	10/10
3	upstash	95.1	0.43s	0.58s	0.58s	10/10
4	archil	94.4	0.45s	0.72s	0.72s	10/10
5	blaxel	94.4	0.53s	0.60s	0.60s	10/10
6	e2b	93.9	0.52s	0.74s	0.74s	10/10
7	vercel	92.7	0.60s	0.93s	0.93s	10/10
8	hopx	83.5	1.54s	1.80s	1.80s	10/10
9	namespace	82.0	1.71s	1.94s	1.94s	10/10
10	modal	78.3	1.89s	2.58s	2.58s	10/10
11	cloudflare	73.7	2.13s	3.39s	3.39s	10/10
12	runloop	73.0	2.04s	3.69s	3.69s	10/10
13	codesandbox	72.7	2.49s	3.08s	3.08s	10/10

Staggered

#	Provider	Score	Median TTI	P95	P99	Status
1	declaw	99.5	0.05s	0.06s	0.06s	10/10
2	daytona	96.6	0.17s	0.58s	0.58s	10/10
3	archil	96.3	0.30s	0.47s	0.47s	10/10
4	upstash	95.8	0.41s	0.43s	0.43s	10/10
5	blaxel	94.6	0.52s	0.57s	0.57s	10/10
6	e2b	94.2	0.50s	0.70s	0.70s	10/10
7	vercel	91.1	0.59s	1.33s	1.33s	10/10
8	hopx	82.5	1.42s	2.26s	2.26s	10/10
9	namespace	80.5	1.83s	2.12s	2.12s	10/10
10	modal	78.5	1.89s	2.54s	2.54s	10/10
11	codesandbox	75.2	2.30s	2.75s	2.75s	10/10
12	runloop	73.8	1.99s	3.57s	3.57s	10/10
13	cloudflare	73.7	2.28s	3.16s	3.16s	10/10

Burst

#	Provider	Score	Median TTI	P95	P99	Status
1	declaw	98.4	0.07s	0.30s	0.30s	10/10
2	daytona	97.4	0.21s	0.35s	0.35s	10/10
3	archil	95.3	0.45s	0.50s	0.50s	10/10
4	upstash	94.8	0.44s	0.65s	0.65s	10/10
5	e2b	94.2	0.48s	0.72s	0.72s	10/10
6	blaxel	94.1	0.55s	0.66s	0.66s	10/10
7	vercel	93.3	0.58s	0.80s	0.80s	10/10
8	hopx	79.9	1.99s	2.03s	2.03s	10/10
9	modal	79.8	1.87s	2.25s	2.25s	10/10
10	namespace	79.6	1.96s	2.16s	2.16s	10/10
11	runloop	75.6	2.04s	3.04s	3.04s	10/10
12	cloudflare	68.5	2.22s	4.53s	4.53s	10/10
13	codesandbox	66.9	3.07s	3.66s	3.66s	10/10

---

View full run · SVGs available as build artifacts

[Copilot]

Pull request overview

This PR aims to improve benchmark reliability by detecting potential sandbox/container reuse across iterations and concurrent launches.

Changes:

• Adds a per-run “reuse detector” (nonce + remembered identity signals) for sequential, concurrent, and staggered benchmark modes.
• Introduces an in-sandbox identity probe that writes marker files and collects runtime/namespace signals to flag suspected reuse.
• Threads the reuse detector through runIteration so different benchmark runners share a single detector per run.

Reviewed changes

Copilot reviewed 3 out of 3 changed files in this pull request and generated 3 comments.

File	Description
src/sandbox/staggered.ts	Creates a per-run reuse detector and passes it into runIteration for staggered launches.
src/sandbox/concurrent.ts	Creates a per-run reuse detector and passes it into runIteration for concurrent launches.
src/sandbox/benchmark.ts	Implements reuse detection: run nonce + identity probe parsing + heuristics, and extends runIteration to accept the detector.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[github-actions]

Browser Benchmark Results

#	Provider	Score	Create	Connect	Navigate	Release	Total	Status
1	Browserbase	94.2	0.22s	0.11s	0.13s	0.14s	0.63s	10/10
2	Hyperbrowser	93.8	0.14s	0.18s	0.10s	0.18s	0.61s	10/10
3	Kernel	91.3	0.08s	0.47s	0.15s	0.07s	0.80s	10/10

---

View full run · SVG available as build artifact

[github-actions]

Storage Benchmark Results

1MB Files

#	Provider	Score	Download	Throughput	Upload	Status
1	Cloudflare R2	94.8	0.11s	76.0 Mbps	0.19s	1000/1000
2	AWS S3	94.7	0.15s	55.6 Mbps	0.06s	1000/1000
3	Tigris	93.7	0.44s	19.0 Mbps	0.21s	1000/1000

4MB Files

#	Provider	Score	Download	Throughput	Upload	Status
1	Cloudflare R2	95.0	0.17s	194.3 Mbps	0.33s	1000/1000
2	AWS S3	92.7	0.81s	41.3 Mbps	0.24s	1000/1000
3	Tigris	91.6	1.07s	31.4 Mbps	0.44s	1000/1000

10MB Files

#	Provider	Score	Download	Throughput	Upload	Status
1	Cloudflare R2	94.3	0.33s	251.5 Mbps	0.73s	1000/1000
2	AWS S3	91.1	1.08s	77.3 Mbps	0.44s	1000/1000
3	Tigris	84.6	3.64s	23.1 Mbps	1.02s	1000/1000

16MB Files

#	Provider	Score	Download	Throughput	Upload	Status
1	Cloudflare R2	94.0	0.49s	272.2 Mbps	0.79s	1000/1000
2	AWS S3	85.7	3.35s	40.1 Mbps	0.50s	1000/1000
3	Tigris	81.4	4.59s	29.2 Mbps	1.06s	1000/1000

---

View full run · SVGs available as build artifacts