OpenAI Devrel

OpenAI Devrel is OpenAI’s developer-relations function, associated with the developer communications account @OpenAIDevs and public events around its models. OpenAI’s API help page directs developers to that account for API feature news and developer-community events. Its public activity spans an o1 release meetup in 2024 and announcements about Codex and API performance in 2026.[3][1][2]

OpenAI DevRel announced Codex integration into Xcode 26.3 through @OpenAIDevs, as reported in a February 4, 2026 Latent Space roundup. The account also announced that GPT‑5.2 and GPT‑5.2‑Codex were 40% faster for API customers, according to the same roundup, which characterized the change as “same weights, lower latency.” The reported improvement concerned API service performance.[2]

Earlier public activity supplies context for this communications role: in September 2024, OpenAI DevRel appeared alongside the Strawberry team at a meetup covering the newly released o1 models.[1]

OpenAI DevRel participated alongside the Strawberry team in an o1 release meetup featured in the September 13, 2024 Latent Space episode. The meetup recap covered the launch of o1-preview and o1-mini. Latent Space described o1’s central change as longer inference-time reasoning: the models spent more computation on step-by-step responses.[1]

The API-platform portion featured Michelle Pokrass, identified at the time as OpenAI’s API Platform leader and a co-leader of the August 2024 Structured Output mode. It traced the progression from function calling in June 2023 to schema-less JSON Mode in November 2023 and then Structured Outputs. That product work belonged to the API-platform discussion, distinct from the episode’s DevRel and Strawberry-team meetup.[1]

By December 2025, OpenAI’s associated Developer Experience work included supporting founders and builders integrating AI into applications and directing developer-tool creation. The December 3 OpenAI Forum event identified Romain Huet as Head of Developer Experience.[4]

Organized by OpenAI Forum, the event advertised a conversation and Codex demonstration with Huet and OpenAI software engineer Aaron Friel. Its program focused on Friel’s use of Codex for multi-step projects, managing complexity through prompts and structure. The event credited Simon Willison with the term “vibe engineering,” framing it as a more rigorous approach to LLM-assisted development in which experienced engineers remain responsible for the resulting software.[4]