# OpenAI Agents API (slim)

> The OpenAI Agents API runs OpenAI's managed Codex harness as a hosted service. One call creates a session with a model, tools, MCP servers, subagents and an OpenAI-hosted or self-hosted sandbox, and OpenAI runs the loop.

- Full: https://www.anchorterminal.com/tools/openai-agents-api.md (~7,800 tokens) · this version ~1,680 tokens · JSON https://www.anchorterminal.com/tools/openai-agents-api.json · canonical https://www.anchorterminal.com/tools/openai-agents-api
- Index: https://www.anchorterminal.com/llms.txt · API: https://www.anchorterminal.com/api/v1/index.json · MCP: https://www.anchorterminal.com/mcp · Updated: 2026-11-06 · pricing https://www.anchorterminal.com/tools/openai-agents-api#pricing · reviews https://www.anchorterminal.com/tools/openai-agents-api#reviews
- Licence: CC BY 4.0, credit Anchor Terminal and link this page (https://www.anchorterminal.com/terms/#data)

**BB · 73.3/100 · rank #99 of 1354 · #2 in Agent harnesses · agent-ready · confidence medium**


Assessment: OpenAI's hosted agent runtime, the Codex harness behind one session call, with MCP, functions, web search, computer use, subagents and sandboxes on OpenAI or your own infrastructure. Documentation, errors and SDKs are thorough. It's still public beta, can't use Zero Data Retention and keeps data in the United States only.

## Facts

- Kind: HTTP API · vendor: OpenAI · category: Agent harnesses · legal entity: OpenAI OpCo, LLC · provenance 95/100
- Endpoint: `https://api.openai.com/v1/agents/sessions` (https)
- Auth: API key · pricing: Pay per use · x402: no · licence: Proprietary hosted API under OpenAI's Services Agreement. The SDKs are Apache 2.0
- Probe metrics: not measured yet (probes haven't run)
- Surface graded: The hosted API at api.openai.com/v1/agents, read from the launch post, the Agents API overview, errors and OpenAI-hosted sandbox guides, the changelog, the pricing page and the status feed on 6 November 2026. Platform readings (keys, SLA, certifications, DPA, subprocessors, deprecations) are the openai-api listing's of 1 November. No account was made and nothing was called. The Agents SDK is the separate openai-agents-sdk listing, and Codex the openai-codex listing
- Sessions: A session holds the agent (model, tools, subagent settings), an environment and the task input. You send input, follow progress by streaming or webhooks, and continue or steer. OpenAI manages orchestration, context compaction and recovery, and keeps session state until you delete it
- Environments: OpenAI-hosted sandboxes in small (1 vCPU, 1 GB), medium (2 vCPU, 4 GB, the default) and large (4 vCPU, 16 GB) sizes, deleted after an hour without activity or keep-alives. Outbound network enabled by default, or disabled, or restricted to 1 to 100 exact host names. Self-hosted sandboxes with workspace and capability directories, and provider guides for AWS Lambda MicroVMs, Blaxel, Cloudflare, Databricks, Daytona, DigitalOcean, E2B, Modal, OCI and Vercel. AWS's Bedrock Managed Agents is built on it
- Tools: MCP servers over HTTP, custom functions, programmatic tool calling, web search, computer use in a hosted browser (since 29 September), plugins, vaults and skills, with automatic review for approvals
- Errors: HTTP codes from 400 (invalid_request_error, invalid_beta and others) to 503 (server_is_overloaded), turn codes such as context_length_exceeded, session_budget_exceeded and rate_limit_exceeded, and session codes such as idle_timeout. Retry guidance says to check the session first, honour Retry-After, back off with a cap, and send only unfinished work after a failed turn
- Data: Session state is retained until deleted, and sessions and published artifacts can be deleted. Not eligible for Zero Data Retention, and residency is United States only. The data-controls guide lists /v1/agents among the endpoints that aren't ZDR-eligible, per the openai-api listing's 1 November reading

- Scores: Reliability 60, Performance pending, Schema & documentation 88, Agent ergonomics 80, Security & auth 87, Payments & pricing 30, Task success pending, Maintenance & community 87, Transparency & trust 85 · total over the 7 assessed categories
- Why: Reliability, Read as a hosted API. · Schema & documentation, The reference is linked from the overview and the SDKs carry beta.agents types; we didn't confirm /agents/sessions in OpenAI's public OpenAP… · Agent ergonomics, Read as an API an agent drives. · Security & auth, Project keys with restricted per-endpoint permissions, service accounts and workload identity, per the openai-api listing; we didn't confirm… · Payments & pricing, No machine payment protocol (0). · Maintenance & community, Closed API. · Transparency & trust, Closed service with clear terms, Apache-2.0 SDKs (15).
- Sources: 9, open questions: 4, both in the full twin
- Capabilities: agent.harness
- JSON: https://www.anchorterminal.com/api/v1/tools/openai-agents-api.json
- Verify (for the vendor): the badge `https://www.anchorterminal.com/badges/openai-agents-api.svg` or a link to https://www.anchorterminal.com/tools/openai-agents-api from a page on openai.com or one of its subdomains, then `POST https://www.anchorterminal.com/api/v1/verify` `{"slug", "url"}` or `verify_listing` at /mcp; re-checked weekly, no effect on the grade. Snippets in the full twin.
- Support: Discord https://discord.com/invite/openai; https://help.openai.com/en/; https://help.openai.com/en/articles/6614161-how-can-i-contact-support; more in the full twin

## Before you call it

1. POST https://api.openai.com/v1/agents/sessions with a project key and the header OpenAI-Beta: agents=v1, or use client.beta.agents in the SDKs
2. Before retrying, retrieve the session and turn and check which actions completed, since there's no idempotency key
3. Set the sandbox network to restricted with an allowlist of up to 100 hosts when the agent doesn't need the open web
4. Save files from /workspace/outputs before deleting a session; no retention period is stated for artifacts
5. Don't send data that needs Zero Data Retention or non-US residency; the Agents API supports neither

## Connect

```bash
curl https://api.openai.com/v1/agents/sessions \
  -H "Authorization: Bearer $OPENAI_API_KEY" \
  -H "OpenAI-Beta: agents=v1" \
  -H "Content-Type: application/json" \
  -d @session.json   # model, tools, environment and input, per the quickstart
```
Full config and headless snippets are in the full page. Through letme (picks today, calling later): https://letme.dev/openai-agents-api

## Similar tools

| Tool | Grade | Score | Shared capabilities | Slim |
| --- | --- | --- | --- | --- |
| OpenAI Codex | BB | 73.5 | agent.harness | https://www.anchorterminal.com/tools/openai-codex.min.md |
| goose | BB | 71.3 | agent.harness | https://www.anchorterminal.com/tools/goose.min.md |
| AXLR | BB | 70.6 | agent.harness | https://www.anchorterminal.com/tools/axlr.min.md |
| OpenHands | BB | 70.3 | agent.harness | https://www.anchorterminal.com/tools/openhands.min.md |
| OpenCode | B | 68.8 | agent.harness | https://www.anchorterminal.com/tools/opencode.min.md |


## Panel reviews (0, desk reviews from public material, no calls made)