
OpenAI Codex
situationalOpenAI's coding agent — terminal CLI, IDE, and cloud tasks, bundled into ChatGPT plans.
Visit chatgpt.com →> **OpenAI Codex** — situational > Verified 2026-08-26 > Limitation: Local and cloud usage share a rolling 5-hour window; weekly caps also apply. > > https://noemium.com/tools/openai-codex/Add to kitOpen in NoemiumFact-check this catalog entry for OpenAI Codex (coding). Verdict: situational Verdict text: The CLI on GitHub is real and the ChatGPT bundle makes it the lowest-friction OpenAI coding agent. Caps sit on a five-hour window, credits replaced per-message billing in Apr 2026, and cloud tasks are Plus-and-up. Not a second Responses API card — this is the product you run against a repo. No field run here, so this is not a Ship stamp. Limitations: - Local and cloud usage share a rolling 5-hour window; weekly caps also apply. - Free/Go are trial-level; cloud GitHub review and plugins sit on Plus+. - Token-credit metering (from 2026-04) makes a session's cost hard to predict. - Conversation/repo context still leaves OpenAI; no self-host of the hosted agent. Receipts: https://chatgpt.com/codex/pricing/ https://github.com/openai/codex https://help.openai.com/en/articles/20001106-codex-rate-card Last verified: 2026-08-26 Find outdated prices, wrong claims, or missing limitations. Cite primary sources (docs, pricing pages, benchmarks, model cards). Do not use marketing pages.[](https://noemium.com/tools/openai-codex/)
ledger entry · observed by @whysanesanders
The CLI on GitHub is real and the ChatGPT bundle makes it the lowest-friction OpenAI coding agent. Caps sit on a five-hour window, credits replaced per-message billing in Apr 2026, and cloud tasks are Plus-and-up. Not a second Responses API card — this is the product you run against a repo. No field run here, so this is not a Ship stamp.
Field guide: OpenAI Codex →
Strengths
- Lowest-friction OpenAI coding agent if you already pay for ChatGPT.
- Real CLI on GitHub plus IDE and cloud tasks on one bundle.
- Free/Go exist as a trial path; Plus is where cloud GitHub review lives.
Known limitations
- Local and cloud usage share a rolling 5-hour window; weekly caps also apply.
- Free/Go are trial-level; cloud GitHub review and plugins sit on Plus+.
- Token-credit metering (from 2026-04) makes a session's cost hard to predict.
- Conversation/repo context still leaves OpenAI; no self-host of the hosted agent.
Use it when
- Repo work on an OpenAI seat you already have, not a second Responses API card.
- Mixed local CLI and cloud tasks inside the ChatGPT plan.
Skip it when
- You need a predictable session cost — credits and a 5-hour window fight you.
- The repo cannot leave OpenAI's hosted agent.
- Multi-step agent work is the job and you can run Cursor or Claude Code instead.
Buy-check
- ·Price — freemium, included in ChatGPT Free/Go/Plus/Pro; Plus $20/mo, Pro 5x $100/mo or 20x $200/mo; extra credits after 5h caps; API key billed at API rates as of 2026-08-26. Open the pricing page, not a screenshot.
- ·Free tier — Free tier exists — burn it first.
- ·Cancel fallout — What happens to your data/projects on cancel — check the ToS, not the FAQ.
- ·Seat math — Per-seat pricing multiplies quietly. Count seats before the annual plan.
- ·Who skips it — Also skipped when: You need a predictable session cost — credits and a 5-hour window fight you.; The repo cannot leave OpenAI's hosted agent..
Price trail — 1 revisionchangelog →
- 2026-W35included in ChatGPT Free/Go/Plus/Pro; Plus $20/mo, Pr…included in ChatGPT Free/Go/Plus/Pro; Plus $20/mo, Pr…
Facts
- pricing
- freemium1
- price note
- included in ChatGPT Free/Go/Plus/Pro; Plus $20/mo, Pro 5x $100/mo or 20x $200/mo; extra credits after 5h caps; API key billed at API rates
- free tier
- yes
- open source
- yes
- api
- yes
- self-host
- no
- category
- coding
- evidence
- source-verified
- Models
- GPT-5.6 SolGPT-5.6 TerraGPT-5.6 Lunabundled
Sources
- 1chatgpt.com/codex/pricing/[snapshot]
- 2github.com/openai/codex[snapshot]
- 3help.openai.com/en/articles/20001106-codex-rate-card[snapshot]
No source, no number.
Similar joball alternatives →
Compare:OpenAI Codex vs Jules
| tool | verdict | price | oss | self-host |
|---|---|---|---|---|
| OpenAI Codex | situational | freemium | yes | no |
| Claude Code | ship | paid | no | no |
| Cursor | ship | freemium | no | no |
| Aider | ship | free | yes | yes |
Facts from the catalog files — not scores you can buy. Quality and speed stay in the briefing, not in a fake 1–5 grid.
Claude Codeanchor
Anthropic's terminal agent that edits, runs and tests code in your repo.
API
paidincluded in Claude Pro $20/mo ($17/mo yearly) and Max 5x $100 / Max 20x $200; shares the plan usage pool; extra usage at API rates
Cursoranchor
AI-first code editor (VS Code fork) with agent mode and inline edits.
FREE TIER
freemiumHobby free; Pro $20/mo; Pro+ $60/mo (3× Agent limits); Ultra $200/mo (20×); Teams from $40/user/mo
Aider
Open-source pair programming in the terminal — git-native, works with any model.
OSSAPIFREE TIER
freefree, open source; bring your own API key