GPT-5.3-Codex
OpenAI's self-improving Codex generation — merges the Codex and GPT-5 training stacks, ~25% faster.
About
GPT-5.3-Codex merges OpenAI's two previously separate lineages — the Codex family (code-generation training stack) and the GPT-5 family (general frontier intelligence) — into one model aimed at steerable, long-horizon agentic coding, with runs stretching past 24 hours and mid-task steering. OpenAI claims ~25% faster inference than GPT-5.2 Codex (provider-reported). The launch narrative leaned hard on self-improvement — 'helped build itself' — which reads as marketing for standard AI-assisted development workflows until independently verified, and some coverage flagged exactly that skepticism. Dating caveat: one detailed report dates the launch to February 4, 2026 (same week as a dedicated Codex Mac app), while a secondary tracker lists June 2026 — the exact day is not firmly verified, so it is left null here.
Specs
Context window
Pricing
Free quota
API access
Weights
Released
Benchmarks
- ~25% faster inference than GPT-5.2 Codex — provider-reported
Provider-reported unless a source says otherwise. Benchmarks are marketing-adjacent by nature — treat them as directional.
Use cases
Trust score — 7/10
Scored 7/10 (metrics tier, metrics-v1). Evidence: Dating conflict: AICLoudIT reports Feb 4, 2026 launch, secondary tracker says June 2026 — releasedAt left null, conflict noted in entry; ~25% faster than GPT-5.2 Codex provider-reported; 'helped build itself' treated as marketing, flagged skeptically in entry; Thinner sourcing than GPT-5-Codex (two sources).