Skip to main content
Codex+55 Strong positive

GILD Review

GILD's call and the buying facts behind it.

GILD analysis

Codex Review

Choose Codex to govern a mixed fleet from one console, at the cost of OpenAI-only models on ChatGPT plans and uneven results on large or ambiguous tasks. Its sandbox starts on and runs natively on Windows, and Agent Security overrides policy per user, group or role on Business and Enterprise. Practitioners rate its code; it pays off on tightly scoped, reviewed tasks, under usage limits that vary by model.

  • AI Coding Agents
  • Engineering Productivity

The decision

Best fit
Teams that delegate well-scoped coding tasks across a mixed fleet, native Windows included, under per-group policy.
Consider another workflow
If you want models from several providers under one plan, or agent work managed entirely inside GitHub, an editor or platform-native tool fits better.
Adoption condition
Enforce sandbox and approval defaults through managed configuration. Commit an AGENTS.md to each repository. Review every change before merge.

Basis: Codex documentation, including its sandbox, policy and pricing pages, plus public accounts from engineers. On the independent Terminal-Bench 4.0 leaderboard of Oct 23, 2026, Codex with GPT-6.1 Sol scores 58.18%.

Buying & deployment factsWhat engineering leaders should verify before a broader rollout.
How you buy it

US$20

A month in ChatGPT Plus. Business is US$25 per user monthly, or US$20 billed annually.

Source 1, OpenAI, Pricing: Pricing

Details

Codex comes with ChatGPT plans. Plus US$20 per month; Pro US$100, US$200, or US$500 per month; Business US$25 per user per month billed monthly, or US$20 billed annually for 2+ users, with Premium seats at US$125, or US$100 billed annually, for five times Standard usage and no five-hour limit; Enterprise through OpenAI sales. Pro 200 subscriptions now carry a lower allowance than before; earlier subscribers kept the old one for a grace period that has ended. Free, and Go at US$8 per month, get Codex in the desktop app on GPT-6 Luna as it rolls out.

Cost drivers

15 to 160 messages

Local messages per five hours on Plus and Business. Pro has no five-hour limit.

Source 1, OpenAI, Pricing: PricingSource 5, OpenAI, OpenAI API pricing: OpenAI API pricingSource 6, developers.openai.com, GPT-6.1 Sol model page: GPT-6.1 Sol model page

Details

Local messages and cloud chats share plan limits, which vary by model. Plus and Business estimate 15 to 160 local messages per five hours on GPT-6.1 Sol, and Pro has no five-hour limit. Past the limit, ChatGPT credits extend usage. With an API key, gpt-6.1-sol costs US$2 input, US$0.10 cached input, and US$10 output per million tokens. Prompts over 272,000 input tokens bill at 2x input and 1.5x output for the whole request.

Where it runs

App, CLI, IDE, cloud

The desktop app, the CLI, editor extensions, and cloud tasks from GitHub, Slack, or Linear.

Source 4, OpenAI, Codex Cloud: Codex CloudSource 7, learn.chatgpt.com, Codex in your IDE: Codex in your IDE

Details

The ChatGPT desktop app, with Windows and Linux desktop apps, the Codex CLI, and an IDE extension for VS Code, Cursor, and Windsurf. Codex Cloud tasks start from the web, mobile, or desktop app, with GitHub, Slack, and Linear integrations.

What to verify

Sign-in route

ChatGPT sign-in follows workspace rules. An API key has no cloud features.

Source 2, OpenAI, Authentication: AuthenticationSource 3, OpenAI, Agent approvals & security: Agent approvals & securitySource 8, OpenAI, Enterprise admin setup: Enterprise admin setupSource 9, GitHub, Codex CLI 0.164.0 release (Oct 23, 2026): Codex CLI 0.164.0 release (Oct 23, 2026)

Details

The sign-in route: ChatGPT sign-in follows workspace RBAC and retention, and Codex cloud requires ChatGPT sign-in, so an API key covers local work only. Check sandbox and network defaults, and that Codex Cloud stays off on Enterprise until an admin enables it. Windows fleets behind an authenticating corporate proxy, or with enterprise MCP servers, need Codex CLI 0.164.0 (Oct 23, 2026) or later. That release adds Negotiate/NTLM proxy sign-in and enterprise MCP authentication.

Sourced from Codex documentation