OpenAI · LIVEBENCH 2026-06-25

GPT-6 Astra Max Effort

Published benchmark results for this specific model configuration.

gpt-6-astra-max

Overall score82.2 / 100

Cost / successful task$0.736USD · benchmark workload

Weight accessNot reported

Where this configuration performs

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

Reasoning92.7

Coding80.4

Agentic Coding57.3

Mathematics96.8

Data Analysis83.0

Language89.4

Instruction Following75.6

Task-level results

theory of mind84.6zebra puzzle100.0spatial98.0logic with navigation88.0code generation80.3code completion80.4javascript63.6typescript43.3python65.0AMPS Hard98.0integrals with game100.0math comp97.1olympiad92.2consecutive events90.6tablejoin58.4tablereformat100.0connections100.0plot unscrambling86.3typos82.0paraphrase78.3simplify70.5story generation74.0summarize79.5

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