← Models

GPT-5.3-Codex

OpenAI · GPT-5

Specialized GPT-5.3 Codex model for long-horizon agentic software engineering.

Part of GPT family · Other versions: GPT-5.5 Pro , GPT-5.5 , GPT-Rosalind , GPT-5.4 mini , GPT-5.4 nano , GPT-5.4-Pro , GPT-5.4 , GPT-5.3 , GPT-5.2-Codex , GPT-5.2-Pro , GPT-5.2 , GPT-5-Codex , GPT-5 mini , GPT-5 , GPT-5 nano , GPT-4.1 , GPT-4o mini Transcribe , GPT-4o mini TTS , GPT-4o Transcribe , GPT-4o mini , GPT-4o
Type
language
Context
400K tokens
Max Output
128K tokens
Status
current
Input
$1.75/1M tok
Output
$14/1M tok
API Access
Yes
License
proprietary
coding agentic software-engineering tool-use code-review refactoring
Released February 2026 · Updated April 18, 2026

Overview

Freshness note: Model capabilities, limits, and pricing can change quickly. This profile is a point-in-time snapshot last verified on April 18, 2026.

GPT-5.3-Codex is still an important OpenAI coding-specialized model, but it should no longer be described as the single current default Codex route across every OpenAI surface. OpenAI’s current pricing docs still list it explicitly as a specialized Codex model, while other official docs describe Codex model choice as surface- and configuration-dependent.

Capabilities

The model is built for multi-file edits, code review, test-aware implementation loops, and longer agent workflows where reliable tool use matters. OpenAI’s launch materials also position it as stronger on computer-style professional work beyond pure coding, including spreadsheets, decks, docs, and longer-running execution.

Technical Details

OpenAI’s current pricing and launch materials continue to place GPT-5.3-Codex in the live model picture, even though the cleaner dedicated model-card style documentation now centers more on GPT-5.1-Codex and GPT-5.2-Codex. OpenAI’s GPT-5.4 launch also explicitly says GPT-5.4 incorporates the frontier coding capabilities of GPT-5.3-Codex, which makes this model an important reference point for current software engineering workflows.

Pricing & Access

Current published standard API pricing (per 1M tokens):

  • Input: $1.75
  • Cached input: $0.175
  • Output: $14.00

OpenAI’s pricing docs still list GPT-5.3-Codex as a specialized Codex model. Treat this as proof of current availability in OpenAI’s model stack, not proof that it is the default model behind every Codex client.

Best Use Cases

Use GPT-5.3-Codex for repository-level implementation, migration work, CI-adjacent code review automation, and coding agents that need stronger patch reliability and longer-horizon execution than general-purpose model routes.

Comparisons

  • GPT-5.4 (OpenAI): Better frontier general model when coding is only part of the workflow.
  • GPT-5.2-Codex (OpenAI): Cleaner current API-model reference when you want a well-documented Codex route.
  • Claude Code + Claude Opus family (Anthropic): Strong alternative coding stack when Anthropic-native tooling is preferred.