GPT-5.2
Released 2025-12 · hybrid · 400K tokens · 13 benchmarks
Editorial notes
Lanzado 11 diciembre 2025. Primer modelo en alcanzar 100% en AIME 2025. La variante Pro alcanza 93.2% en GPQA Diamond.
Spec sheet
- Company
- OpenAI
- Country
- US
- Type
- hybrid
- Release
- 2025-12
- Context
- 400K tokens
- Pricing (openai)
- $1.25/$5/M
- Slug
- gpt-5-2
Benchmarks (13)
Reasoning 4
- 92.4GPQA-DiamondGraduate-level Physics, Chemistry, Biology — PhD-level questions.
- 91.4MMLUMassive Multitask Language Understanding — 57 academic subjects, ~16K questions.
- 75.4MMLU-ProMMLU upgraded with harder questions and 10 answer options.
- 52.9ARC-AGI-2Updated ARC challenge — novel and tough abstract reasoning.
Coding 3
Math 2
Knowledge 1
Multilingual 1
Cite this model
BibTeX · APA
BibTeX
@misc{frontier-gpt-5-2,
title = {GPT-5.2},
author = {{OpenAI}},
year = {2025},
note = {Frontier Benchmarks AI atlas. Accessed 2026-05-08},
url = {https://frontierbenchmarks.com/models/gpt-5-2}
} APA
OpenAI (2025). GPT-5.2 [Large language model]. Frontier Benchmarks AI. Retrieved 2026-05-08, from https://frontierbenchmarks.com/models/gpt-5-2
Citation reflects the atlas page, not the original model paper. For the paper, see the "Resources" section above.