Global storyplatform-engineer

Pay per second only for the compute a sandbox actually uses

A global story is meaningful for any software product, so it can be compared across the whole site — every product’s verdict on per-second-billing across all 2 arenas that carry it, judged from public evidence.

7/13 products pass (full or partial) · click a verdict for the product’s rationale and evidence

Adoption among tracked products

53.8%

Pay per second only for the compute a sandbox actually uses: 7 of 13 tracked products pass (full or partial), across 2 arenas.

Tracking since Sep 2026 — the diffusion curve appears once a second month of tracking accrues.

Honest approximation: what this curve can and can’t show

We don’t have per-verdict change history, so the curve shows adoption among products as they enter tracking (first score-history entry), with each product carrying its current verdict — not the moment each product shipped the capability. Exact verdict history accrues from here forward.

Product	Arena	Verdict	Quality	Evidence
Runpod	GPU Clouds	full	9/10	5 sources
Modal Sandboxes	Agent Sandboxes & Code Execution	full	8/10	4 sources
Lambda	GPU Clouds	full	8/10	2 sources
Daytona	Agent Sandboxes & Code Execution	partial	6/10	2 sources
E2B	Agent Sandboxes & Code Execution	partial	4/10	6 sources
Vercel Sandbox	Agent Sandboxes & Code Execution	partial	4/10	4 sources
Blaxel	Agent Sandboxes & Code Execution	partial	3/10	1 source
Cloudflare Sandboxes	Agent Sandboxes & Code Execution	none	—	3 sources
CoreWeave	GPU Clouds	none	—	3 sources
Maritime	Agent Sandboxes & Code Execution	none	—	2 sources
Paperspace	GPU Clouds	none	—	2 sources
Runloop	Agent Sandboxes & Code Execution	none	—	0 sources
Vast.ai	GPU Clouds	none	—	0 sources

Verdicts are judged per arena against that arena’s evidence packs, so the same tier can rest on different evidence depth across arenas — follow a row to the product page for the full rationale.