Vendor — 04 active of 4 cataloged # OpenAI [openai.com ↗](https://openai.com/)/`vndr://openai`/verified 2026-09-25 OpenAI ships 4 cataloged agents — CLI agents, computer-use agents, research agents, agent communication platforms — over cli · editor · api · chat · browser interfaces with local + cloud hosting. RUBRIC ## Portfolio spec sheet | Agent | Category | Interfaces | Hosting | Pricing | Status | Caps | | --- | --- | --- | --- | --- | --- | --- | | Codex CLICodex is OpenAI's coding agent: an open-source terminal CLI, a desktop app, IDE extensions, and a cloud lane for async tasks, all included in ChatGPT plans. | CLI agent | cli · editor · api | local · cloud | Included in ChatGPT Free/Go/Plus/Pro/Business/Enterprise; credits for extra usage | active | 8 tags | | ChatGPT AgentChatGPT Agent is OpenAI's hosted computer-use mode — the Operator lineage: it spins up a managed browser, works through multi-step web tasks, and can take actions after confirmation. | computer-use agent | chat · browser | cloud | Included in ChatGPT Plus/Pro/Business; task quotas by tier | active | 5 tags | | ChatGPT Deep ResearchChatGPT Deep Research is OpenAI's research-agent mode: a prompt becomes a 5–30 minute browsing run that returns a cited report — now also reachable via API as a deep-research model. | research agent | chat · api | cloud | Included in ChatGPT tiers (query limits vary); API via deep-research models | active | 4 tags | | OpenAI Agents SDKThe OpenAI Agents SDK is OpenAI's lightweight framework for agentic apps — its handoffs primitive is a first-class agent-to-agent mechanism: one agent transfers the live conversation to a specialist. | agent communication platform | api | local · cloud | Free (OSS SDK); model usage metered | active | 4 tags | Capability tags — the atomic taxonomy. Open an agent for the full list. NOTES ## Field notes [Codex CLI](https://www.agentlist.io/systems/codex-cli) OpenAI-native terminal and async coding when you already pay for ChatGPT. Caveat — Included usage varies by ChatGPT plan; extra work is billed as credits and rates differ by model. Check the Codex rate card. [ChatGPT Agent](https://www.agentlist.io/systems/chatgpt-agent) Individuals who want a capable general-purpose web agent with zero setup — research, bookings, spreadsheet-and-browser chores. Caveat — Task quotas are tier-bound; it will still ask you to take over on logins, payments, and CAPTCHAs — the 'autonomous' part is deliberately bounded. [ChatGPT Deep Research](https://www.agentlist.io/systems/deep-research) The default deep-research baseline — thorough public-web reports inside the subscription most people already pay for. Caveat — Query limits are tier-bound; it still misattributes claims occasionally, and paywalled/niche sources are largely out of reach. [OpenAI Agents SDK](https://www.agentlist.io/systems/openai-agents-sdk) Teams on OpenAI models who want minimal scaffolding for supervisor→specialist delegation without a heavyweight framework. Caveat — OpenAI-first (other providers via community adapters); handoffs transfer a single conversation rather than providing a general messaging fabric. CAP ## Capability coverage - [API-native](https://www.agentlist.io/capabilities/api-native) - [Background runs](https://www.agentlist.io/capabilities/background-tasks) - [Browser operation](https://www.agentlist.io/capabilities/browser-ops) - [Cited answers](https://www.agentlist.io/capabilities/citations) - [Code execution](https://www.agentlist.io/capabilities/code-exec) - [Desktop control](https://www.agentlist.io/capabilities/desktop-use) - [File editing](https://www.agentlist.io/capabilities/file-edits) - [Git & PRs](https://www.agentlist.io/capabilities/git-pr) - [Human handoff](https://www.agentlist.io/capabilities/human-handoff) - [Long-horizon tasks](https://www.agentlist.io/capabilities/long-horizon) - [MCP tools](https://www.agentlist.io/capabilities/mcp) - [Multi-agent orchestration](https://www.agentlist.io/capabilities/multi-agent) - [Repo context](https://www.agentlist.io/capabilities/repo-context) - [Shell access](https://www.agentlist.io/capabilities/shell) - [Tool calling](https://www.agentlist.io/capabilities/tool-calling) - [Web research](https://www.agentlist.io/capabilities/web-research) VS ## Head-to-head (3) - [→Claude Code vs Codex CLI](https://www.agentlist.io/compare/claude-code-vs-codex-cli) - [→ChatGPT Agent vs Manus](https://www.agentlist.io/compare/chatgpt-agent-vs-manus) - [→ChatGPT Deep Research vs Gemini Deep Research](https://www.agentlist.io/compare/deep-research-vs-gemini-deep-research) PLUG ## Integrations (8) - [Chrome / Chromium](https://www.agentlist.io/integrations/chrome) - [GitHub](https://www.agentlist.io/integrations/github) - [Gmail](https://www.agentlist.io/integrations/gmail) - [Google Drive](https://www.agentlist.io/integrations/google-drive) - [MCP servers](https://www.agentlist.io/integrations/mcp-servers) - [SharePoint](https://www.agentlist.io/integrations/sharepoint) - [Slack](https://www.agentlist.io/integrations/slack) - [VS Code](https://www.agentlist.io/integrations/vscode) WIRE ## Transports (3) - [Email (SMTP/IMAP)](https://www.agentlist.io/transports/email) - [MCP (Model Context Protocol)](https://www.agentlist.io/transports/mcp) - [Slack](https://www.agentlist.io/transports/slack) PEER ## Vendors in the same categories - [Community / OSS9 agents](https://www.agentlist.io/ai/community) - [Google6 agents](https://www.agentlist.io/ai/google) - [Anthropic2 agents](https://www.agentlist.io/ai/anthropic) - [Browserbase2 agents](https://www.agentlist.io/ai/browserbase) - [xAI2 agents](https://www.agentlist.io/ai/xai) - [A2A Project1 agent](https://www.agentlist.io/ai/a2a-project) - [AGNTCY1 agent](https://www.agentlist.io/ai/agntcy) - [Alibaba Cloud1 agent](https://www.agentlist.io/ai/alibaba) Vendor pages are generated from the catalog seed. Coverage reflects systems cataloged as of 2026-09-25; it is not a claim about a company's full product line. --- Source: [OpenAI AI — 4 agents cataloged | agentlist.io](https://www.agentlist.io/ai/openai). This is the public page rendered as Markdown; interactive controls require the website.