Google · LIVEBENCH 2026-06-25

Gemini 3.8 Flash High

Published benchmark results for this specific model configuration.

gemini-3.8-flash-high

Overall score75.8 / 100

Cost / successful task$0.307USD · benchmark workload

Weight accessNot reported

Where this configuration performs

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

Reasoning89.3

Coding72.5

Agentic Coding54.2

Mathematics91.6

Data Analysis54.0

Language87.8

Instruction Following81.4

Task-level results

theory of mind76.9zebra puzzle98.3spatial100.0logic with navigation82.0code generation73.2code completion71.7javascript72.7typescript40.0python50.0AMPS Hard99.0integrals with game80.0math comp95.1olympiad92.2consecutive events20.6tablejoin45.3tablereformat96.1connections100.0plot unscrambling73.4typos90.0paraphrase80.0simplify75.4story generation82.8summarize87.5

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