LuminaBench
Google / Published benchmarks

Gemini 3.1 Pro (Preview)

Benchmark results and API pricing.

Best recorded results

Benchmark scores

Best eligible score in each benchmark.

ARC-AGI-21,095Not specified · ExperimentalARC-AGI-3962Not specified · ExperimentalGPQA Diamond1,090High · ExperimentalMystery Game Puzzles1,048High · ExperimentalLumina Reasoning Index1,092Not specified · 1/3 Provisional · Experimental

API pricing

USD / 1M tokens

Verified pricing is unavailable for this model.

ARC-AGI-2

ARC-AGI-2

Version 2
Lumina VerifiedLumina checked the published sources, versions, model labels, score extraction and conversion. External evaluations were run by their publishers; Lumina has not independently rerun or reproduced them.

1 configuration · Publisher-run · Experimental conversion

1,095Best score · Not specified
Explore ARC-AGI-2
ARC-AGI-3

ARC-AGI-3

Version 3
Lumina VerifiedLumina checked the published sources, versions, model labels, score extraction and conversion. External evaluations were run by their publishers; Lumina has not independently rerun or reproduced them.

1 configuration · Publisher-run · Experimental conversion

962Best score · Not specified
Explore ARC-AGI-3
GPQA Diamond

GPQA Diamond

Version Diamond-198
Lumina VerifiedLumina checked the published sources, versions, model labels, score extraction and conversion. External evaluations were run by their publishers; Lumina has not independently rerun or reproduced them.

2 configurations · Publisher-run · Experimental conversion

1,090Best score · High
Explore GPQA Diamond
Mystery Game Puzzles

Mystery Game Puzzles

Version Puzzles-100
Lumina VerifiedLumina checked the published sources, versions, model labels, score extraction and conversion. External evaluations were run by their publishers; Lumina has not independently rerun or reproduced them.

3 configurations · Publisher-run · Experimental conversion

1,048Best score · High
Explore Mystery Game Puzzles

Lumina Reasoning Index

Version 3
By LuminaIndex designed by LuminaBench using attributed, publisher-run benchmarks. Lumina did not rerun those evaluations.

5 configurations · Derived from published benchmarks · Experimental · 1/3 Provisional

1,092Provisional score · Not specified
Explore Lumina Reasoning Index