Term
GPT-5.2
GPT-5.2 is OpenAI's model generation from December 2025 — three variants (instant, thinking, Pro), a 400k context window and new state-of-the-art on professional knowledge work.
GPT-5.2 — explained in more detail
OpenAI released GPT-5.2 on December 11, 2025, roughly three weeks after Google’s Gemini 3 Pro. The generation ships in three variants: GPT-5.2 instant for fast replies, GPT-5.2 thinking as the reasoning model with standard and extended thinking modes, and GPT-5.2 Pro with higher reasoning effort and correspondingly more compute. Both 5.2 models offer a 400,000-token context window with up to 128,000 output tokens. Knowledge cutoff: August 31, 2025.
Example / Practical use
GPT-5.2 Thinking set a new state-of-the-art score at launch on GDPval — an eval measuring knowledge work across 44 occupations. It was the first model to perform at or above the level of human experts, beating or tying top professionals on 70.9% of comparisons. Alongside it, OpenAI shipped a coding-specialized variant, GPT-5.2-Codex. Improvement themes: long-context understanding, agentic tool-calling, vision, plus building spreadsheets and presentations.
Entdecke mehr
Three Days with Claude Fable 5 — and Why Opus Stays My Daily Driver
Fable 5 fixed a bug in one shot that Opus failed at twice. I'm still not switching. My honest impressions from the trenches after three days.
GlossarGPT-4o
GPT-4o (May 2024) was OpenAI's first natively multimodal model — text, audio and image in a single model, with a real-time voice mode and substantially lower cost than GPT-4 Turbo.
LexikonComparing AI models — who builds what and how to choose
The major model families in 2026 at a glance. Who builds Claude, GPT, Gemini, Llama, Mistral, DeepSeek, Qwen — and which model to pick when.