Agent leaderboards / All sectors / File storage
File storage: which products coding agents choose
Amazon S3 won about 46% of runs, and led for every agent.
Read this leaderboard as textrankings, key learnings, method
Key learnings
We asked three coding agents to add file storage to 8 small apps, 90 runs in all. We asked in different words, and as four different people. Amazon S3 came first, and the rest of the runs went to Azure Blob Storage, Google Cloud Storage, Cloudflare R2 and Tigris.
Who asked changed the answer
In the 36 runs as enterprise teams, Azure Blob Storage and Google Cloud Storage split the work evenly, 18 each. Senior engineers went to Amazon S3 in 20 of their 27 runs.
Second place depended on the agent
Claude Code put Cloudflare R2 second, picking it eight times in 30 runs. Cursor and Codex both had Azure Blob Storage second, with six each.
The wording moved a few answers
We counted a case as one app with one agent, asked over and over in new words. Five of the 24 cases came back with more than one product. The rest gave the same answer every time.
- The simulated user approved all 90 plans, and sent the agent back in only two runs.
- In three runs it refused to approve until the agent named a specific product.
- Tigris won once, in the SvelteKit indie SaaS note-taking app.
The ranking90 runs
| Product | Wins | Share | ||
|---|---|---|---|---|
| 1 | Amazon S3aws.amazon.com | 41 | 46% | |
| 2 | Azure Blob Storageazure.microsoft.com | 18 | 20% | |
| 3 | Google Cloud Storagecloud.google.com | 18 | 20% | |
| 4 | Cloudflare R2cloudflare.com | 12 | 13% | |
| 5 | Tigristigrisdata.com | 1 | 1% |
By agent, by persona, by wording
By agent
| Cursor · Grok 4.630 runs | Amazon S3 · 15then Azure Blob Storage · 6 |
| Claude Code · Claude Opus 530 runs | Amazon S3 · 10then Cloudflare R2 · 8 |
| Codex · GPT-5.6 Sol30 runs | Amazon S3 · 16then Azure Blob Storage · 6 |
By persona
| Enterprise team36 runs | Azure Blob Storage · 18then Google Cloud Storage · 18 |
| Senior engineer27 runs | Amazon S3 · 20then Cloudflare R2 · 7 |
| Junior developer18 runs | Amazon S3 · 13then Cloudflare R2 · 5 |
| Vibe coder9 runs | Amazon S3 · 8then Tigris · 1 |
By what the ask stressed
| The plain ask90 runs | Amazon S3 · 41then Azure Blob Storage · 18 |
A case is one codebase with one agent, asked several times in different words and as different people. 5 of 24 cases did not hold to a single product.
How this was measured
Every number on this page comes from a controlled experiment. We took 8 small applications, asked 3 coding agents (Cursor (Grok 4.6), Claude Code (Claude Opus 5), Codex (GPT-5.6 Sol)) to add file storage to each of them, in several wordings and as a enterprise team and senior engineer and junior developer and vibe coder, and let the agent choose the product. Each run happened in a sandbox with the agent at a pinned version, and a judge read the session to record what was chosen. That is 90 runs. The interactive board shows every run with its session, its diff and the judge's verdict. A simulated user stood in for the owner of the codebase: it read the agent's plan and had to approve it before any code was written; it sent the agent back at least once in 2 runs. Read the methodology and the publications.
If you sell in this sector: what these numbers mean for a vendor.
Open the interactive boardThis page as Markdown