OpenAI · LIVEBENCH 2026-11-06
gpt-6-astra
Published benchmark results for this specific model configuration.
gpt-6-astraOverall score85.8 / 100
Cost / successful task$5.233USD · benchmark workload
Weight accessNot reported
Where this configuration performs
Category averages from the same release, on a 0–100 scale.
Task-level results
long agentic75.6reasoning92.7coding87.5browser91.3data analysis83.0research90.6design80.0
Source files: Scores ↗ Categories ↗ Costs ↗ . Missing cost is unknown, never free. Compare configurations within the same release.