GPT-5 Codex

by OpenAI

GPT-5 Codex is a version of GPT-5 optimized for agentic coding, trained with reinforcement learning on real-world coding tasks across diverse environments. It generates code that closely mirrors human style and pull-request preferences, follows instructions precisely, and iteratively runs tests until it achieves passing results. With a 272k token context window, GPT-5 Codex supports structured output and tool use for integration with development workflows. It is built for developers who need reliable code generation with human-like conventions and testing loops baked into the process. As part of the GPT-5 family's evolution in coding, GPT-5 Codex serves as a foundation for agentic coding workflows, with newer variants such as GPT-5.2 Codex and GPT-5.3 Codex offering stronger performance on large-scale code transformations and complex software engineering.

Key info

Input
Output
Features
Context window
272K
Max output
128K

Benchmarks

Independent benchmark scores — composite indices for reasoning, coding, and math, plus individual eval scores where available.

Intelligence Index37.0
Math Index98.7
Reasoning & knowledge
MMLU-Pro
87%
GPQA Diamond
84%
Humanity's Last Exam
28%
Long-context reasoning
71%
Coding
LiveCodeBench
84%
SciCode
41%
Agentic & tool use
Terminal-Bench Hard
38%
τ²-Bench Telecom
87%
Math & instruction following
AIME 2025
99%
IFBench
74%

Available routes

No routes currently available — GPT-5 Codex isn't routed through the Opper gateway right now. It may return.

Contact us about this model →

Available models from OpenAI

Start building with 700+ models

One API key. Every major provider. Up and running in minutes.

Get startedView Documentation
GPT-5 Codex by OpenAI — not currently on Opper | Opper AI