gpt-5.3-codex

Upstream reference: GPT-5.3-Codex

Publisher: OpenAI

OpenAI documents GPT-5.3-Codex for tool-using software-engineering workflows. The figures below describe the published model reference, not a measured Sigma5C runtime configuration.

Publisher-described context, separate from Sigma5C observations and recorded catalog statistics.

Published upstream specifications

Published context window

400,000 tokens Source

Published maximum output

128,000 tokens Source

Publisher-described uses

• Publisher-described tool-using software-engineering workflows. Source

Reference limitations

• The official page marks this model deprecated; account access and replacement policy are separate from historical observations. Source
• An alias, client or account can impose different access and configured limits. Source

A recorded tag and an upstream card association do not verify the deployed checkpoint, quantization, configured context or current hosted availability.

Publisher sources

• GPT-5.3-Codex official documentation — OpenAI. Checked 2026-11-02.

Links

Recorded references do not establish downloadable weights, licensing or current availability.

• OpenAI — official site
• Provider page
• Official model announcement / documentation

Where gpt-5.3-codex comes from

Named in 2 observation pages

• Codex CLI — 8 OpenAI models head-to-head · V1
• Codex vs Claude — native audit of a 425K LOC project · V1