Do coding agents recommend OpenAI models?
OpenAI models was chosen in 4% of 623 judged document processing sessions, ranking sixth.
OpenAI models was chosen in 4% of 623 judged document processing sessions, ranking sixth. It was also raised as a candidate in 308 further sessions without being chosen.
This page reports what happened when Claude Code, Codex, Cursor, Grok Build CLI and Muse Code had to solve a problem in document processing inside a realistic codebase. Not what a chat assistant says about OpenAI models. What an agent actually installed.
The numbers
| Category | Document processing |
| Sessions in the category | 623 |
| Sessions where OpenAI models was chosen | 22 |
| Install share | 4% |
| Rank in category | 6 of 40 |
| Codebases it won in | 6 |
| Raised as a candidate, not chosen | 308 |
| Chosen when considered | 7% |
| Site | openai.com |
By agent
The agents treat OpenAI models very differently. Muse Code chose it in 19% of its sessions in this category and Claude Code in 0%, a spread of 19 points.
| Agent | Sessions | Chose OpenAI models | Share |
|---|---|---|---|
| Claude Code | 180 | 0 | 0% |
| Codex | 175 | 6 | 3% |
| Cursor | 180 | 4 | 2% |
| Grok Build CLI | 30 | 1 | 3% |
| Muse Code | 58 | 11 | 19% |
A single blended install share would report the midpoint and hide both ends. If most of your users run Claude Code, your real position here is 0%, not 4%.
By who was asking
OpenAI models performs similarly across the four kinds of buyer, from 0% to 9%. That is unusual: the category leader changed with the persona in 23 of the 29 categories we measured.
| Who is asking | Sessions | Chose OpenAI models | Share |
|---|---|---|---|
| Vibe coder | 53 | 5 | 9% |
| Junior developer | 267 | 14 | 5% |
| Senior engineer | 211 | 3 | 1% |
| Enterprise team | 92 | 0 | 0% |
What OpenAI models was up against
The full ranking in document processing, from the same sessions:
| # | Product | Runs won | Share |
|---|---|---|---|
| 1 | Anthropic Claude | 136 | 22% |
| 2 | Azure AI Document Intelligence | 136 | 22% |
| 3 | Amazon Textract | 77 | 12% |
| 4 | Google Cloud Document AI | 44 | 7% |
| 5 | Google Gemini | 39 | 6% |
| 6 | OpenAI models (this page) | 22 | 4% |
| 7 | Ocrolus | 19 | 3% |
| 8 | Apache PDFBox | 18 | 3% |
What this means
This is an integration problem, not a presence problem. Agents raised OpenAI models in 308 sessions and chose it in 22, so it reaches the shortlist and then loses. Something at the last step is costing the session, and in our data that is usually a quickstart that does not run when pasted, documentation describing an interface that changed, or a package name that does not match the product name.
That is the cheaper of the two problems to have. The reason is written down in each losing transcript.
Where these numbers come from
The 623 sessions in document processing are part of a published set of 11,878, run with real coding agents inside realistic codebases and judged blind. The full method is on one page: how we measured this.
Every document processing run can be replayed on the board.
If you work on OpenAI models: the judge recorded a reason for every session where it was raised and passed over. Those reasons are in the transcripts.
<!-- generated by scripts/write-data-pages.mjs -->
Common questions
Do coding agents recommend OpenAI models?
Yes. OpenAI models was chosen in 22 of the 623 judged sessions in document processing, a 4% install share, ranking sixth in its category.
Does Claude Code recommend OpenAI models?
In 0 of the 180 sessions in document processing run with Claude Code, which is 0%.
Do different coding agents treat OpenAI models differently?
Yes, and by a wide margin. Muse Code chose it in 19% of its runs and Claude Code in 0%.
How was this measured?
Real coding agents at pinned versions were run in sandboxes inside 91 realistic codebases and asked to solve real tasks. A simulated project owner approved or questioned each recommendation before any code was written, and a judge from a model family that builds none of the agents read every session blind.
How often is OpenAI models considered but not chosen?
It was raised as a candidate in 308 sessions without being chosen, and chosen in 22. That is a 7% conversion from considered to chosen.
Where this comes from
Armature ran 11,878 judged sessions with Claude Code, Codex, Cursor, Grok Build CLI and Muse Code inside 91 realistic codebases, and published every run. The numbers on this page come from that work.