The Model Index
Up to four models

Gemini 3 Pro vs Gemini 2.5 Pro (Jun 2025)

On the index (GPT-4 = 100), Gemini 3 Pro scores 160 and Gemini 2.5 Pro (Jun 2025) 145. Side by side on the same third-party tests, with price and context. The link in your address bar shares this exact comparison.

Next to Gemini 3 Pro
Presets
Gemini 3 Pro
160145–176
WeightsClosed
ReleasedNov 2025
Price in / out–
Context–
Core tests taken5 of 9
Gemini 2.5 Pro (Jun 2025)
145129–161
WeightsClosed
ReleasedJun 2025
Price in / out$1.25 / $10
Context1.05M
Core tests taken5 of 9

Gemini 3 Pro leads Gemini 2.5 Pro (Jun 2025) by 15 points, ahead on 4 of 4 shared core tests.

Test by testthird-party results

Gemini 3 ProGemini 2.5 Pro (Jun 2025)
0%25%50%75%100%REASONING & KNOWLEDGEGPQA Diamond (Epoch AI)GPQA Diamond (Epoch AI)Humanity's Last ExamHumanity's Last ExamGeoBenchGeoBenchSimpleBenchSimpleBenchVPCTVPCTMATHSAIME-style maths (OTIS mock)AIME-style maths (OTIS mock)FrontierMath (2025 set)FrontierMath (2025 set)FrontierMath tier 4 (2025 set)FrontierMath tier 4 (2025 set)CODINGSWE-bench VerifiedSWE-bench VerifiedTerminal-Bench 2.0 (Terminus 2 runs)Terminal-Bench 2.0 (Terminus 2 runs)GSO-BenchGSO-BenchWeirdMLWeirdMLAGENTSDeepResearch BenchDeepResearch Bench
Every result as a table
TestGemini 3 ProGemini 2.5 Pro (Jun 2025)
Reasoning & knowledge
GPQA Diamond (Epoch AI)92.6%85.3%
Humanity's Last Exam37.5%21.6%
GeoBench84.0%81.0%
SimpleBench76.4%62.4%
VPCT91.0%46.4%
Maths
AIME-style maths (OTIS mock)91.4%84.7%
FrontierMath (2025 set)37.6%14.1%
FrontierMath tier 4 (2025 set)18.8%4.2%
Coding
SWE-bench Verified72.9%57.6%
Terminal-Bench 2.0 (Terminus 2 runs)56.9%32.6%
GSO-Bench18.6%3.9%
WeirdML69.9%54.0%
Agents
DeepResearch Bench46.3%42.8%
Task length · METR43 sec33 sec

Tests at least two of these models have taken. Bold rows are the core tests.

This page as a chartDownload PNG