{"slug": "openai-redrew-one-web-page-our-parser-read-zero-models", "title": "OpenAI Redrew One Web Page. Our Parser Read Zero Models.", "summary": "A developer built a general-purpose fallback reader to replace hand-written parsers that silently returned zero models after OpenAI redesigned its deprecations page on Nov 12, 2026, and tested it across every provider page before shipping. The reader was wrong in seven distinct ways on other providers' pages that never surfaced on OpenAI's, so the team built a comparison harness that deliberately stales stored rows and verifies exactly those rows are corrected. After an independent review returned 19 findings (18 fixed with tests, one accepted as a stated limit), the change went to production on Nov 13, reading 166 models on OpenAI's redrawn page and moving 16 changed replacements while withdrawing nothing.", "body_md": "On Nov 12, 2026, OpenAI redesigned its deprecations page. Where there used to be a section per notice, there was now one table of upcoming deprecations.\n\nThe next time our parser read it, it found 0 models.\n\nNot \"fewer models\". Zero. A page that had been feeding us retirement dates and replacements was, as far as our code could tell, empty.\n\nSince tracks when AI models are retired on each platform, among a lot of other things. Those dates come from the providers' own pages, and for each page we had a hand-written parser that knew its layout.\n\nWhen a page changes shape, a parser like that does not crash. It just reads nothing. That is the dangerous case, because \"I found no deprecated models\" and \"every model was removed\" are one bad assumption apart.\n\nWe had a guard for exactly this. A page that suddenly reads far below its usual count is refused, and everything we knew from it stays as last read. So nothing was deleted and no dates vanished.\n\nThe guard did its job. It also meant that everything we knew about OpenAI was now frozen.\n\nFrozen is fine for a day. It is not fine when the page is still changing underneath you.\n\nEvery scan showed OpenAI's source as failing. And the replacements went stale: OpenAI had changed the named replacement for `gpt-5-mini` to `gpt-6.1-terra`, and we were still saying `gpt-5.6-terra`. Nothing about that is loud. It is just wrong, quietly, on every answer that touches the Dec 11 shutdown.\n\nThe obvious move was to fix the OpenAI parser for the new table. Then wait for the next redesign and fix it again.\n\nOur first plan was better than that, but only a little. When a page's own parser reads too little, hand the page to a general reader that reads it the way a person would, find the model, its date and its replacement, and keep only what the page proves.\n\nWe built it and tested it on OpenAI's page. It worked.\n\nThat is where I pushed back. OpenAI is not special. Every provider redraws its pages, all the time. A reader that is only tested on the one page that broke this week is tested on the one page we already know about.\n\nSo we ran it against every page it could ever be given.\n\nRun on the other providers' pages, live, the same reader that was right on OpenAI was wrong in seven different ways:\n\nNone of those show up on OpenAI's page. Every one of them would have shipped if we had stopped at the page that broke.\n\nWhat fixed it was not a smarter reader. It was building the comparison first and tuning second.\n\nFor every page, the test went like this:\n\nThen the harder half: make some stored rows stale on purpose, a wrong date here, an old replacement there, run it again, and check that exactly those rows get put right. Not more, not fewer.\n\nWe also cached the reader's answers by input, so changing the code around it cost nothing to re-run.\n\nThe other change was about trust. A general reader is flexible, which is the point, and also the risk. So it got a strict job description:\n\nThat second line is the one I would put on a poster. A reader that cannot find something has not proven it is gone.\n\nBefore release, an independent review went through the change and came back with 19 findings. All 19 were real. 18 were fixed, each with a test. The last one is a stated limit: a page read this way never withdraws a date.\n\nTwo of the 19 mattered most. A version's row on Foundry would have been read as the bare alias stored beside it, and production has those rows; our first per-page test had missed it. And a date the old parser could not read, like \"December 2026\", would have withdrawn a retirement that was still standing.\n\nAfter the fixes, on Anthropic, Gemini, Vertex AI, Microsoft Foundry, Bedrock and Mistral: 0 rows moved, 0 rows added, nobody told, and every row we broke on purpose came back right. On OpenAI's redrawn page, end to end in our test environment: the source was active again, 166 models read, the 16 replacements that had changed were moved, and nothing was withdrawn.\n\nIt went to production on Nov 13 at 5:40 PM MST.\n\nThe same week, two providers moved dates. Microsoft Foundry pushed `claude-haiku-4-5` and `claude-opus-4-5` out to Feb 22, 2027. Google pushed its Gemini text-to-speech previews from Nov 17 to Dec 18, 2026.\n\nThat is the whole point, really. These pages are not documents you read once. They are live data that someone redesigns whenever they feel like it.\n\nIf you read one page you control, a hand-written parser and a loud failure are simpler and better. A general reader earns its place when the pages belong to someone else, there are many of them, and they change without telling you. Even then it only works with the boring parts around it: the harness, the proof rule, and the guard that refuses a page that suddenly reads as empty.\n\nThe thing I took from it is smaller than all that. When a fix works on the case that broke, that is the moment to go find every other case it will meet, before the fix meets them in production.\n\nWhat's the worst thing a page you scrape changed without warning, and how long did it take you to notice?\n\n*I'm Nabil. I build [Since](https://since.dev?utm_source=devto), which tracks what your code depends on and tells you before it breaks.*", "url": "https://wpnews.pro/news/openai-redrew-one-web-page-our-parser-read-zero-models", "canonical_source": "https://dev.to/nabil_since_dev/openai-redrew-one-web-page-our-parser-read-zero-models-37ck", "published_at": "2026-11-18 23:30:00+00:00", "updated_at": "2026-11-18 23:41:56.098271+00:00", "lang": "en", "topics": ["ai-tools", "ai-infrastructure", "mlops", "developer-tools", "large-language-models"], "entities": ["OpenAI", "Anthropic", "Google Gemini", "Vertex AI", "Microsoft Foundry", "Amazon Bedrock", "Mistral", "gpt-5-mini"], "also_reported_by": [], "takeaway": "A general LLM reader used as a fallback for brittle page parsers must be validated against every provider's page, not just the one that broke, and must never withdraw a stored fact it simply failed to find.", "recommendation": {"advice": "When a scraper or reader returns suspiciously few results, refuse the update and keep last-known data; then test the fallback across all sources with deliberately staled rows to confirm exactly those rows are corrected.", "playbook": "output-verification", "playbook_url": "https://wpnews.pro/playbooks/output-verification"}, "alternates": {"html": "https://wpnews.pro/news/openai-redrew-one-web-page-our-parser-read-zero-models", "markdown": "https://wpnews.pro/news/openai-redrew-one-web-page-our-parser-read-zero-models.md", "text": "https://wpnews.pro/news/openai-redrew-one-web-page-our-parser-read-zero-models.txt", "jsonld": "https://wpnews.pro/news/openai-redrew-one-web-page-our-parser-read-zero-models.jsonld"}}