OpenAI status — and how to poll it yourself

At this snapshot, 8 of the 9 providers on our status board reported fully operational. Here is where OpenAI stood, and how to get the same answer from its status API without a browser.

The 34 components OpenAI reports status for

When OpenAI says "operational", this is the granularity behind the claim — the component list its own status feed publishes, fetched 2026-08-31. All were reported operational at snapshot time. If an incident hits, these are the names its status page will scope it to.

• Ads API
• Ads Manager
• Agent
• Audio
• Batch
• CLI
• Chat Completions
• ChatGPT Atlas
• ChatGPT Work
• Codex API
• Codex Web
• Codex in ChatGPT Desktop
• Compliance API
• Connectors/Apps
• Conversations
• Deep Research
• Embeddings
• FedRAMP
• File uploads
• Files
• Fine-tuning
• GPTs
• Image Generation
• Images
• Login
• Login
• Moderations
• Realtime
• Responses
• Search
• Sites
• Sora
• VS Code extension
• Voice mode

Polling OpenAI's status programmatically

Atlassian Statuspage v2. GET /api/v2/status.json returns status.indicator (none | minor | major | critical) and a human description; GET /api/v2/components.json returns per-component state.

curl -s https://status.openai.com/api/v2/status.json

Everything above was verified against the live endpoint on 2026-08-31 — nothing is inferred from platform defaults. If the endpoint moved, tell us.

Model retirements — OpenAI

OpenAI accounts for 80 of the 129 announced retirements in our deprecations table: 36 dated retirements still ahead as of 2026-08-31, 43 already past. 1 more row carries hedged wording (e.g. “at earliest 2024-06-13”) kept verbatim from the provider's docs. Every row cites the provider's own deprecations page.

Full table: all 129 rows, filterable, with CSV/JSON download →

Other providers