OpenAI · closed · 2026-02

GPT-5.3-Codex

Item: GPT-5.3-Codex
Rating: 79.1

not ranked n/a

the take
The Codex-line coding specialist, now being quietly retired in favour of the general models. Good SWE-bench, thin on everything else, which is why it sits here rather than on the board.

Not ranked: OpenAI doesn't publish enough comparable public benchmarks to place it fairly. How ranking works.

benchmarks

80.0

—

—

—

LiveCodeBench (coding)

77.3

specs

Context: 400K
Input: $1.75/M
Output: $14/M
Speed: —
Modality: text, image

strengths

Strong agentic coding (~80% SWE-bench)
~25% faster than GPT-5.2-Codex
Mid-tier pricing

weaknesses

Being superseded by GPT-5.5/5.6
Coding scores largely OpenAI-reported
Thin on non-coding benchmarks

Sources: [1][2]

Back to leaderboard Battle GPT-5.3-Codex