OpenAI

OpenAI is referenced throughout the wiki as a model provider, API platform, and agentic product company. It appears as the source of the Responses API, Codex, Workspace Agents, Whisper, GPT Image 2, GPT-5.4, GPT 5.5, and the realtime audio models Gpt Realtime Translate, Gpt Realtime Whisper, and Gpt Realtime 2.

Key facts

• Responses API integrated in n8n OpenAI chat model node v1.3+
• Built-in tools: web search (with domain, location, context-size filters), file search, code interpreter, MCP
• Vector store file search: $0.10/GB/day, charged even when unused
• GPT-5 Mini required for web search with domain restrictions (not available on 4.1)
• Extra options via responses API: conversation ID (server-side memory), prompt cache key, safety identifier, service tier, saved prompts, metadata, top log probs
• In ^[src-042], GPT-style public API pricing is discussed as an observable signal of underlying serving costs: output-token premiums, cache-write/cache-hit prices, and long-context tiers can reveal decode and memory-bandwidth constraints.
• In ^[src-048], OpenAI's Codex is presented as a full ChatGPT-adjacent agentic app with project workspaces, model/reasoning controls, plugins, file access, browser automation, skills, scheduled automations, and deployment workflows.
• In ^[src-051], OpenAI demonstrates realtime audio models in the API: Gpt Realtime Translate for live translation across 70 languages and Gpt Realtime 2 for reasoning voice agents with context preservation and tool actions.
• In ^[src-058], OpenAI's Codex is presented as a multi-interface work app with plugin connectors, custom plugin creation, project folders, Codex automations, Chronicle screen memory, and image generation inside software projects.
• In ^[src-061], OpenAI is framed as chaotic but unusually good at landing defining research-product moves such as Deep Research, Sora, and o1-style thinking models; GPT-5's router is also discussed as a likely way to reduce GPU spend by routing many users away from expensive high-reasoning paths.
• In ^[src-081], OpenAI presents Codex as an everyday-work agent for researchers, educators, operators, small-business owners, leaders, and knowledge workers, not only developers.
• The same source surfaces OpenAI's trust posture for enterprise agents: sandboxed execution, scoped file access, optional network restrictions, read-only modes, enterprise controls, and auto-review agents that supervise risky actions.
• In ^[src-083], OpenAI presents the realtime audio stack as three models: Translate for multilingual speech translation, Realtime Whisper for streaming transcription, and GPT Realtime 2 for reasoning voice agents with 128k context, preambles, parallel tool calls, and voice-to-action application control.
• The same source uses Sierra as a production example, emphasizing that enterprise voice agents need harnesses around the model: workflows, tools, grounding, guardrails, VAD, traces, redaction, payment controls, simulations, and supervision.
• In ^[src-084], OpenAI's product direction is Codex-centered: GPT-5.4, tool search, hosted shell, code mode, Chrome control, Workspace Agents, prompt caching, and harness engineering all push toward agents that work across code, browsers, tools, schedules, and team workflows.
• The same cluster adds OpenAI's policy frame through Sam Altman: superintelligence debate should begin early, and society may need AI-enabled resilience, broad compute access, labor-transition mechanisms, and new institutions.
• In ^[src-104], OpenAI's GPT Realtime 2 is used as the voice layer for a local desktop agent, showing how the API can be paired with push-to-talk capture, browser automation, MCP-connected apps, and accessibility-tree desktop control.
• In ^[src-112], OpenAI's Codex advanced configuration docs become a local-AI monitoring signal because Robin flagged the OSS mode and local-provider section for future reference.
• In ^[src-113], OpenAI's 2026-06-16 Codex app changelog is tracked as official product-change evidence for Codex's evolution into a broader agentic work surface.
• In ^[src-118], Financial Times reporting adds a market-economics signal: OpenAI's frontier race is tied to large 2025 spending, revenue, R&D, marketing, and loss figures ahead of planned IPO activity.

Source references

• ^[src-005] Nate Herk cluster — Nate Herk — n8n cluster (18 videos)

– Videos referenced: lokbsA5VXOk, jx-yvSRLKrA, 5p5cV0yVDvQ

• ^[src-042] Dwarkesh Patel — "How GPT, Claude, and Gemini are actually trained and served – Reiner Pope" (2026-04-29)
• ^[src-048] Nate Herk — "Master 97% of Codex in 1 Hour (full course)" (2026-05-06)
• ^[src-051] OpenAI – "We’re introducing three audio models in the API" (2026-05-07)
• ^[src-058] Jack Roberts — "How to use Codex Better than 99% of People" (2026-05-06)
• ^[src-061] Lex Fridman – "State of AI in 2026: LLMs, Coding, Scaling Laws, China, Agents, GPUs, AGI | Lex Fridman Podcast #490" (2026-01-31)
• ^[src-081] OpenAI — "Codex for Everyday Work: AI Agents Beyond Coding" (2026-05-14)
• ^[src-083] OpenAI – "Build Hour: GPT-Realtime-2" (2026-05-13)
• ^[src-084] OpenAI Codex, Workspace Agents, Prompt Caching, and Superintelligence Policy cluster (2026-02-09 to 2026-05-08)
• ^[src-104] Pat Simmons – "GPT Realtime 2 Can Now Run Your Entire Computer (Just Your Voice)" (2026-06-17)
• ^[src-112] OpenAI Developers – "Advanced Configuration – Codex" (unknown)
• ^[src-113] OpenAI Developers – "Codex 2026-06-16 app changelog" (2026-06-16)
• ^[src-118] Financial Times – "OpenAI spending hit $34bn last year ahead of planned IPO" (2026-06-16)

2026-06-27 update

• New OpenAI signals span public-health funding ^[src-159], Codex and agentic work research communication ^[src-164], reported release gating for GPT-5.6 ^[src-165], and an official GPT-5.6 Sol bookmark that still needs body extraction ^[src-168].

2026-06-27 agentic work evidence update

• OpenAI's official Codex article and paper add a stronger evidence layer to its agentic-work narrative: adoption is measured through output mix, concurrency, skills, task complexity, and role spread ^{[src-169] [src-170]}.
• The paper is especially useful for AI Productivity Measurement because it warns that OpenAI-internal usage is an unusually favorable frontier environment rather than a typical company baseline ^[src-170].

Additional source references

• ^[src-169] OpenAI – "How agents are transforming work" (2026-06-25)
• ^[src-170] Drew Johnston, David Holtz, Alex Martin Richmond, Christopher Ong, Prasanna Tambe, Aaron Chatterji / OpenAI – "The shift to agentic AI: Evidence from Codex" (2026-06-25)

2026-06-29 GPT-5.6 preview update

• OpenAI's GPT-5.6 preview now has body-level capture rather than only a blocked metadata bookmark: Sol is positioned as the flagship, Terra as a balanced everyday model, and Luna as the fast affordable tier ^[src-168].
• The preview is strategically important because OpenAI links frontier model release, agentic coding/cyber/biological capabilities, and U.S. government coordination in the same announcement ^[src-168].

2026-07-03 EU jobs transition framework update

• OpenAI Economic Research extends its labor-market measurement work to the EU through the AI Jobs Transition Framework, combining technical exposure with human necessity and demand elasticity across ESCO occupations [src-192, src-193].
• The report is an important policy signal because OpenAI frames AI workforce planning around occupational structure, national readiness plans, training systems, worker-flow data, and early-warning indicators rather than a simple job-loss forecast ^[src-193].

EU framework source references

• ^[src-192] OpenAI – "Mapping Europe's AI Workforce Opportunity" (2026-06-29)
• ^[src-193] Alex Martin Richmond / OpenAI Economic Research – "The AI Jobs Transition Framework for the EU" (2026-06)