Agent Rank breakdown
- Autonomy
- 8
- Capabilities
- 4
- Integrations
- 4
- Pricing
- 10
- Polish
- 6
- Verifiability
- 10
Auto-computed from autonomy, capabilities, integrations, pricing, maturity and editorial verification. Updated every deploy. How is this computed?
Capabilities
- Code
- Tool use
Integrations
- Version control
- GitHub
- Terminal / shell
- Shell
Pricing tiers
- +Apache-2.0 binary
- +BYO OpenAI / Anthropic key
- +Composable from Makefiles + scripts
Pros & cons
- +Zero-install terminal workflow — agent where engineers already live
- +Open source under permissive license; auditable end-to-end
- +Ideal for one-shot refactors, audits, and migrations
- −No long-running session UI; not built for hour-long autonomous work
- −Less polished than IDE-native alternatives
- −You wire the model key + handle rate limits
OpenAI’s open-source terminal agent for refactors, audits and migrations. Start with the open source tier.
Frequently asked questions
What is Codex CLI?+
Codex CLI is OpenAI's open-source terminal coding agent. Apache 2.0 license, BYO model API key (typically GPT-5 or any OpenAI-compatible endpoint). Sits in your shell and edits files, runs tests, opens PRs — same shape as Claude Code but OpenAI-built and fully open source.
Codex CLI vs Claude Code?+
Claude Code is bundled with Claude Pro ($20/mo includes tokens) and has the more polished autonomy loop. Codex CLI is open source (Apache 2.0) and uses your own API key — better for OpenAI-stack teams, audit-sensitive industries, or anyone who wants to fork the agent itself.
Does Codex CLI work with local LLMs?+
Yes — Codex CLI works with any OpenAI-compatible endpoint. Point it at Ollama, vLLM, or LM Studio and you can run it fully air-gapped with self-hosted models. This is one of its biggest advantages over Claude Code, whose local-LLM support is more limited.
User reviews
Write a review →Be the first to review. We verify every reviewer by email.
