Gemini 3 Flash Preview

Closed Source

Closed Weights

Parameters

Context Length

1.05M

Modality

Multimodal

Architecture

Dense

License

Proprietary

Release Date

8 Jan 2026

Knowledge Cutoff

Jan 2025

Evaluation Benchmarks

Rank

#11

Benchmark	Score	Rank
Graduate-Level QA GPQA	0.904	🥉 3
Professional Knowledge MMLU Pro	0.89	⭐ 4
Agent Arena Agent Arena	0.11	⭐ 5
StackUnseen ProLLM Stack Unseen	0.83	9
General Text high Text Arena	1490	⭐ 11
General Text Standard Text Arena	1474	29
Web Development WebDev Arena	1438	64

Rankings

Overall Rank

#11

Coding Rank

#12

About Gemini 3 Flash Preview

Gemini 3 Flash Preview is a high-speed multimodal model from Google DeepMind optimized for low latency and high-concurrency production tasks. It supports configurable reasoning depth and native video/audio processing over a 1M token context window.

Technical Specifications

Detailed architecture specifications are limited for proprietary models.

Resources

Official Documentation Release Notes

About Gemini 3

Google's latest generation multimodal models with breakthrough performance across coding, mathematics, reasoning, and language understanding. Features ultra-large context windows, native multimodal processing, and thinking modes with minimal latency overhead. Available in Pro and Flash variants optimized for different workloads, with preview versions showing state-of-the-art results on multiple benchmarks.