← All models

Best Model for Codex CLI

Codex CLI is built around OpenAI's GPT-5 models. GPT-5.3-Codex is the coding-tuned default that powers most local runs; GPT-5.5 is the pick for the hardest reasoning; GPT-5.4 is the value sweet spot; and GPT-5.4 Mini handles cheap, high-volume steps.

Read the full Codex review → · updated 2026-06-22
#1GPT-5.3-CodexOpenAI · $5.42/1M · 400K ctx

The coding-tuned default — built for sandboxed execution, diffs, and long agentic coding sessions in Codex.

Overall 89/100 · Value 47 · 77% community winFull breakdown →
#2GPT-5.5OpenAI · $13/1M · 1.1M ctx

The flagship for the hardest reasoning and the trickiest refactors, when quality outweighs cost.

Overall 93/100 · Value 40 · 79% community winFull breakdown →
#3GPT-5.4OpenAI · $6.25/1M · 1.1M ctx

The value sweet spot — most of 5.5's ability for under half the price, with a 1M context.

Overall 90/100 · Value 46 · 77% community winFull breakdown →
#4GPT-5.4 MiniOpenAI · $1.88/1M · 400K ctx

For routing, classification, and the many small calls in a run where a flagship would be wasteful.

Overall 82/100 · Value 58 · 74% community winFull breakdown →

FAQ

Which model does Codex CLI use by default?
Codex CLI leans on OpenAI's coding-tuned model, GPT-5.3-Codex, for local runs. You can switch to GPT-5.5 for harder tasks or GPT-5.4 Mini for cheap, fast steps.
Can Codex CLI run non-OpenAI models?
Codex CLI is designed for OpenAI's models and is most capable on them. For a model-agnostic terminal agent, Aider, Cline, or OpenCode let you run Claude, Gemini, or open-weight models instead.

The model is half the story — the agent is the other half

The model picks the moves; the agent runs the loop, the tools, and the guardrails. Once you've chosen a model, see which agent gets the most out of it.

Compare AI agents →