Atlas

Models

← All models

Gemini 1.5 Pro (source unspecified)

Google

This benchmark entry preserves generative-model observations labeled Gemini 1.5 Pro or an unversioned provider alias when the source does not identify an exact checkpoint. The Gemini 1.5 Pro family includes distinct February and May research versions and later 001 and 002 API snapshots; publication dates and mutable aliases alone do not resolve an observation to one of them. The stable Atlas slug is retained for these unresolved observations, with no exact release date or context limit asserted.

Benchmark scores

BenchmarkScore
A-Fantasia62.3
A-Fantasia (Valid Attempts with Retries)62.3
A-Fantasia Backwards Spelling88.0
A-Fantasia Chess35.0
A-Fantasia Chess (Valid Attempts with Retries)35.0
A-Fantasia Cube (Valid Attempts with Retries)64.0
A-Fantasia Cube Rotation64.0
A-Fantasia Spelling (Valid Attempts with Retries)88.0
AI2D (OpenVLM)79.1
AidanBench1487
Artificial Analysis Coding Index19.8
Artificial Analysis Progress Index v1150
BuzzBench54.9
Epoch Capabilities Index (ECI)127
ForecastBench Baseline Dataset Brier Index57.7
ForecastBench Baseline Market Brier Index58.6
ForecastBench Baseline Overall Brier Index58.1
ForecastBench Preliminary Dataset Brier Index57.6
ForecastBench Tournament Dataset Brier Index57.6
ForecastBench Tournament Market Brier Index62.8
ForecastBench Tournament Overall Brier Index59.6
FORTRESS Average Risk Score (ARS)53.9
GPQA Diamond (Artificial Analysis avg@5)37.1
HallusionBench (OpenVLM)45.6
Humanity's Last Exam (Text-Only)3.5
LisanBench Difficulty-Weighted120
LisanBench Path Length492
LiveCodeBench (Artificial Analysis)24.4
MATH-500 (Artificial Analysis source)67.3
MATH-Vision19.2
Loading Atlas data…