#22
Gemini 3 Flash
gemini-3-flash-preview

Gemini 3 Flash achieves Pro-level intelligence at Flash-level speed and cost. Scoring 90.4% on GPQA Diamond and 81.2% on MMMU-Pro, it delivers frontier reasoning capabilities with ultra-low latency. The model supports dynamic thinking levels and excels at coding (78% SWE-bench), making it ideal for agentic workflows, real-time applications, and complex document analysis.

Performance Metrics

Helpfulness
Instruction Following
Comprehension
Empathy
Creative Writing
83.46
Helpfulness
88.5
Empathy
76
Instruction Following
90.4
Creative Writing
72
Comprehension
90.4
Speed
Avg 95 tok/s

Model Specifications

Release Date
December 2025
Lab
Google
Type
Proprietary
Context Size
1M
Max Output Tokens
65.5K
Cost per 1M tokens
$0.50 / $3.00
Model Inputs
Text, Audio, Images, Video, PDFs
Model Outputs
Text
Tool Calling
Enabled

Compare With Similar Models