OpenAI · closed · 2025-12

GPT-5.2

Item: GPT-5.2
Rating: 88.7

#2 index 88.7

the take
The value play in the GPT-5 line. Reasoning within a hair of the flagship at less than half the price. Ignore the 100 on AIME, that benchmark is saturated now and tells you nothing about the model.

benchmarks

80.0

92.4

—

100.0

LiveCodeBench (coding)

—

specs

Context: 400K
Input: $1.75/M
Output: $14/M
Speed: 73.8 tok/s
Modality: text, image

strengths

Near-flagship reasoning at <half the price
Strong SWE-bench (~80%)
Mature and broadly available

weaknesses

Superseded on the frontier by 5.4/5.5
Scores largely OpenAI-reported
Smaller 400K context

Sources: [1][2]

Back to leaderboard Battle GPT-5.2