GPT-5.3 Codex

Name: GPT-5.3 Codex
Price: 1.75 USD
Author: OpenAI

GPTv5.3-codexCurrent

OpenAI(ai lab)

Released February 5, 2026

API Docs Playground

Context400K

Price (In / Out)$1.75/M / $14.00/M

Training CutoffAugust 2025

CategoryLarge Language Model

Max Output128K

Capabilities

What GPT-5.3 Codex can do and the modalities it supports.

Features

textvisioncodetool useextended thinkingcomputer use

Input Modalities

textimage

Output Modalities

text

Benchmarks

Performance scores for GPT-5.3 Codex across standard benchmarks.

Benchmark	Score	Date
SWE-Bench Pro	56.8	February 5, 2026
Terminal-Bench 2.0	77.3	February 5, 2026
OSWorld-Verified	64.7	February 5, 2026
MMLU	93	February 5, 2026
MMLU Pro	83	February 5, 2026
GPQA	81	February 5, 2026
IFEval	94	February 5, 2026
SimpleQA	58	February 5, 2026
HLE	36	February 5, 2026

Pricing

Token pricing for GPT-5.3 Codex API usage.

Input Tokens

$1.75/M

per million tokens

Output Tokens

$14.00/M

per million tokens

Available only via Responses API. Supports prompt caching. Batch API pricing available at 50% discount. Configurable reasoning effort: low, medium, high, xhigh.

Competing Models

Alternative models that compete with GPT-5.3 Codex.

Model	Family	Context Window	Price (In / Out per M)	Status
Claude Sonnet 4.6	Claude	200K	$3.00/M / $15.00/M	Current
Gemini 2.5 Pro	Gemini	1.0M	$1.25/M / $10.00/M	Current

About GPT-5.3 Codex

OpenAI's specialized agentic coding model, released February 2026. GPT-5.3-Codex powers OpenAI's Codex agent and is the first model classified as High capability for cybersecurity under OpenAI's Preparedness Framework. It achieves state-of-the-art on SWE-Bench Pro (56.8%) across four programming languages, 77.3% on Terminal-Bench 2.0 for complex agentic coding, and 64.7% on OSWorld-Verified for computer use. The model supports configurable reasoning effort (low, medium, high, xhigh) and excels at the full software lifecycle: debugging, deploying, monitoring, writing PRDs, data analysis, and more. It uses fewer tokens than any prior model while maintaining top performance, reducing costs for iterative development workflows.

Technical Details

API Identifier: gpt-5.3-codex
Category: Large Language Model
Context Window: 400,000 tokens
Max Output Tokens: 128,000 tokens