Model

OpenAI Codex

CLOSED

OpenAI · Codex family · released Aug 10, 2021

GPT-3 fine-tuned on GitHub code; powered the original GitHub Copilot and introduced the HumanEval benchmark; Codex API discontinued 2023-03-23.

ReasoningCodingVisionFunction callingTool useAgentic

14.4

Elo · unrated

Parameters

12B

Active params

12B (dense)

Context

4K tokens

Architecture

Dense decoder-only Transformer (GPT-3 fine-tuned on code)

License

Proprietary

Languages

—

API price (in/out)

No hosted API

Modalities

text

Benchmark results

Bar shows position within the tracked field; marker = field best

HumanEvalCoding28.8%#159

best: Claude Opus 4.5 · 99.4%

Run it locally

Closed weights — available via API only. No local deployment.

Input / M tok

—

Output / M tok

—

Codex family

Elo progression across releases

Resources

API price — · each benchmark row carries its own source badge (see methodology)