Google DeepMind

deepmind.googleUS

Portfolio Stats

Total Models8

Multimodal7

Benchmarks Run40

Avg Performance58.6%

Latest Release

Gemini 3.1 Pro

Released: Feb 19, 2026

Multimodal

Release Timeline

Recent model releases by year

2026

1 model

2025

7 models

Performance Overview

Top models and benchmark performance

Top Performing Models

By avg score

76.0%

74.4%

73.1%

73.0%

64.6%

Benchmark Categories

Knowledge

91.8%

Science

82.5%

Coding

63.8%

Finance

57.5%

Agents

51.3%

Model Statistics

Multimodal Ratio

88%

Models with Providers

All Models

Complete portfolio of 8 models with advanced filtering

		License
#01Gemini 3.1 Pro Google DeepMind's iterative upgrade to Gemini 3 Pro with stronger reasoning and improved token efficiency	Feb 19, 2026	Proprietary	80.6%	-
#02Gemini 3 Flash Google's Gemini 3 Flash with 1M context, multimodal support, and fast agentic coding capabilities	Dec 17, 2025	Proprietary	78.0%	-
#03Gemini 3 Pro Google's Gemini 3 Pro with 1M context, multimodal reasoning, and broad world knowledge for complex tasks	Nov 18, 2025	Proprietary	78.0%	-
#04Gemini 2.5 Flash Google's Gemini 2.5 Flash with 1M context, multimodal input, and hybrid thinking at efficient cost	Jun 17, 2025	Proprietary	-	-
#05Gemini Diffusion Google DeepMind's experimental diffusion-based text model for fast code and text generation	May 20, 2025	Proprietary	-	76.0%
#06Gemini 2.5 Pro Google's Gemini 2.5 Pro with 1M context, multimodal reasoning, and strong coding performance	May 20, 2025	Proprietary	-	-
#07Gemma 3 27B Google DeepMind's 27B open-weight multimodal model for instruction-following and visual reasoning	Mar 12, 2025	Modified MIT	-	74.4%
#08Gemma 3 12B Google DeepMind's 12B open-weight multimodal model with vision support and 128K context	Mar 12, 2025	Modified MIT	-	73.0%

Resources

Official Website