OpenAI · LIVEBENCH 2026-06-25

GPT-5.2 High

Published benchmark results for this specific model configuration.

gpt-5.2-2025-12-11-high

Overall score74.6 / 100

Cost / successful task$0.234USD · benchmark workload

Weight accessNot reported

Where this configuration performs

Category averages from the same release, on a 0–100 scale.

Reasoning83.2

Coding76.1

Agentic Coding50.3

Mathematics93.2

Data Analysis78.2

Language79.8

Instruction Following61.8

Task-level results

theory of mind78.8zebra puzzle84.0spatial96.0logic with navigation74.0code generation76.1code completion76.1javascript59.1typescript36.7python55.0AMPS Hard97.0integrals with game92.0math comp95.1olympiad88.6consecutive events88.9tablejoin45.6tablereformat100.0connections99.0plot unscrambling58.4typos82.0paraphrase64.1simplify54.1story generation65.7summarize63.2

Source files: Scores ↗ Categories ↗ Costs ↗ . Missing cost is unknown, never free. Compare configurations within the same release.