GPT-5.3 Codex

Code-Optimized

gpt-5.3-codex · OpenAI · text

Back to Models Back to Docs

Platform

OpenAI

Visit Official Site

API Pricing

Input / 1M

$1.75/M$1.22/M30% OFF

Output / 1M

$14/M$9.8/M30% OFF

Cache Read / 1M

$0.175/M$0.1225/M30% OFF

Overview

GPT-5.3 Codex is OpenAI's coding-specialized model, optimized for code generation, completion, refactoring, and transpilation. With a 272K context window, it supports FIM (Fill-in-the-Middle) completion, Function Calling, and cached_input. It outperforms GPT-5.2 on coding tasks at the same price.

Capabilities

272K context; cached_input automatic caching (50% input); FIM completion; Function Calling; Structured Output; code execution; JSON mode; Batch API at half price.

Use Cases

Code generation and refactoring; multi-file editing; code transpilation; unit test generation; code review; Agent coding workflows; IDE integration.

Limitations

No image generation; no native cache_write support; performance on non-coding tasks is comparable to GPT-5.2; high concurrency requires a quota increase.

Examples

Claude Code

Anthropic official CLI coding agent. Multi-file edits, tool use, long-context reasoning.

Terminal

export ANTHROPIC_AUTH_TOKEN="YOUR_API_KEY"
export ANTHROPIC_BASE_URL="https://api.robovai.com"
export ANTHROPIC_MODEL="gpt-5.3-codex"
claude

Context Window

272K ctx

Thinking Mode

Not supported

Overview

GPT-5.3 Codex, optimized specifically for programming tasks, with standout long-context code generation and multi-file coordinated editing.