Back to Gap Map Product / UX / Orchestration & agents

Codex CLI

OpenAI
open source / Overall score: 4.0(strong)

Terminal coding agent from OpenAI, written in Rust. It reads a codebase, proposes a plan and executes multi-file changes with sandboxed command execution, and documents writable roots, permission profiles, agent approvals, auto-review, MCP connectors and skills. Desktop builds are offered for Linux and Windows.

Codex runs against either a ChatGPT plan or an API key.

Openness

5 high confidence
5.0
license
Apache-2.0(OSI, the CLI agent)
source
public(GitHub)
note
agent is OSS but optimized for/defaults to OpenAI models (provider-leaning, not license-restricted)
core features withheld
no — one Apache-2.0 LICENSE at the root governs the whole tree

Codex CLI is released under Apache-2.0, and the published Rust source is the whole of the CLI agent. Signing in with a ChatGPT plan or an API key pays for model access rather than for any CLI feature withheld from the source, so nothing about the agent itself is gated.

Adoption

4 high confidence
4.0

OpenAI and independent tracking both report a weekly active developer count for Codex CLI, corroborated by stars on the same GitHub repository. That figure is measured weekly rather than the monthly count this scale is written in, so it likely understates the number a monthly measure would show.

Capability

4 medium confidence
4.0

Capable, heavily-used coding agent backed by frontier OpenAI models, but no standalone agent SWE-bench placement is published, so this reads from its feature set: high, but not the open-agent benchmark frontier-definer (OpenHands 68.4%).

  • https://developers.openai.com/codex/cli recorded 2026-09-29

    The docs page lists Sandboxing, Auto-review and a "subagents" section: "Ask Codex to delegate focused work to specialized agents". Searched for a benchmark figure: none.

Verified 2026-09-29