OpenAI · Codex

Codex-12B

READER SUMMARY

A 12-billion-parameter GPT model fine-tuned on public GitHub code for generating and completing computer programs.

DOOM SCORE31.0out of 100model risk profile, not the overall index

CURRENT ASSESSMENT · REVISION 1

Why this model scores 31.0

Codex-12B showed material code-generation breadth but remained a research checkpoint with low independent agency and limited direct deployment. Code generation creates meaningful misuse and control concerns without implying the distinct production Copilot checkpoint was this exact model.

Capability45

Autonomy4

Deployment18

Misuse potential28

Control difficulty20

0 comments · 0 votesOpen discussion

No comments yet. Start the discussion.

MODEL-ATTRIBUTED EVIDENCE

News tied to Codex-12B

The model score of 31.0 rates this model's risk profile. The overall Doom Index of 60.8 measures the recalibrated complete evidence corpus. These values answer different questions.

NET MODEL-ATTRIBUTED INDEX CONTRIBUTION+0.02

Each article's leave-one-out Doom Index contribution is divided equally among the exact models named on that article. This prevents multi-model evidence from being claimed in full on several model pages. Model risk scores are recalculated from their technical profile and related news, but never feed back into the overall index.

07 Jul 2021CapabilityTOWARD

OpenAI publishes Codex-12B and documents practical coding deployment

OpenAI published Codex-12B as a GPT language model fine-tuned on public GitHub code and documented that a distinct production Codex checkpoint powered GitHub Copilot's coding assistant.

Full item contribution: +0.02
Codex-12B equal share: +0.02

Read assessment →

AUDIT TRAIL

Model score history

R1
Doom Score 31.0
New exact public research model absent from the durable catalogue.
12 Aug 2026