leejk/ jk lee
Models: lining up the frontier base models

Landscape · 2026.06

OpenAI — GPT Models

OpenAI's base models, gathered in one place

·
#models#openai#gpt#landscape

A vendor page collecting OpenAI's base models (the GPT-5.5 line). Per-model strengths, weaknesses, and operational constraints are sourced in each standalone piece; this page only sorts which model takes which job.

TL;DR. A vendor page gathering OpenAI's base models. Per-model strengths, benchmarks, and operational constraints are backed by primary sources in each standalone piece below; here we only classify which model takes which job.

What this slot is

The set of base models OpenAI ships for general use. It is one vendor column in the models slot that coding agents, frameworks, and skills ride on — and even within one vendor the job splits per model.

Models

Each model's strengths and weaknesses are sourced in its standalone piece.

  • GPT-5.5 — strong on agentic/terminal coding (Terminal-Bench 2.0 SOTA), enterprise documents (dominant on OfficeQA Pro), and price ($5/$30, half Fable's input). But Opus 4.7 leads on raw SWE-Bench Pro Public, hallucination is weak, and it holds no solo #1 domain. Previous generation — it handed the flagship slot to GPT-5.6 on 2026-07-09.
  • GPT-5.6 (2026-07-09; limited preview 06-26) — no standalone piece yet. Three variants — Luna, Terra, Sol — with flagship Sol scoring 80 on the AA Coding Agent Index v1.1, ahead of its competitor while using less than half the output tokens at roughly a third of the cost. Sol is also the line OpenAI positions as its strongest cybersecurity model (the GPT-5.6-Cyber derivative followed on 08-10).

Sources

Sub-documents