Major Release Google Multimodal Proprietary

Google Gemini 3 Pro

First Gemini 3 tier release; strong multimodal + long-context.

Released 2025-11-18 · ~1T MoE · 1M tokens context · knowledge cutoff 2025-09

Overview

Google's flagship model with Deep Think mode, ranked #1 on LMSYS Arena at launch

Specifications

DeveloperGoogle
Release date2025-11-18
Model typeMultimodal
Parameters~1T MoE
ArchitectureSparse MoE
Context window1M tokens
Max output32K
Knowledge cutoff2025-09
LicenseProprietary
Input modalitiestext, image, audio, video, PDF
Output modalitiestext

Benchmarks

Benchmark Gemini 3 Pro Gemini 2.5 FlashΔ
ARC-AGI 87.5%
MMLU 93.2% 87.5% +5.7
LMSYS Arena #1
MMLU-Pro 89.4% 82.8% +6.6
MMMU 82.1% 79.7% +2.4
GPQA Diamond 78.5%
SWE-bench Verified 68.2%

Pricing

$2.50
per 1M input tokens
$10.00
per 1M output tokens

Availability

API Only Google AI Studio Vertex AI Gemini API

Official references

Launch coverage