Gemini 3 Flash Preview is a high speed, high value thinking model designed for agentic workflows, multi turn chat, and coding assistance. It delivers near Pro level reasoning and tool use performance with substantially lower latency than larger Gemini variants, making it well suited for interactive development, long running agent loops, and collaborative coding tasks. Compared to Gemini 2.5 Flash, it provides broad quality improvements across reasoning, multimodal understanding, and reliability.
The model supports a 1M token context window and multimodal inputs including text, images, audio, video, and PDFs, with text output. It includes configurable reasoning via thinking levels (minimal, low, medium, high), structured output, tool use, and automatic context caching. Gemini 3 Flash Preview is optimized for users who want strong reasoning and agentic behavior without the cost or latency of full scale frontier models.
Evaluations
13
across 13 benchmarks
Latency
16.6s
Context length
1M
tokens
Cost
$0.50 · $3
input · output per 1M tokens
Key takeaways
Gemini 3 Flash Preview (medium) is a Transformer-based model optimized for agentic workflows, multi-turn chat, and coding assistance, balancing strong reasoning with lower latency and cost.
It supports multimodal inputs (text, images, audio, video, PDFs) for text output, features a 1M token context window, configurable reasoning levels, structured output, and tool use.
The model excels in mathematical problem-solving with 97% accuracy on MATH-500, showing strong capabilities in algebra, applying mathematical formulas, number theory, and combinatorial problems.
Despite its strengths, it struggles with highly intricate mathematical problems or subtle misinterpretations of constraints and shows a low accuracy of 27.07% on the 'Humanity's Last Exam' (Biology/Medicine Subset), indicating difficulty with deep, nuanced, or specialized external knowledge.
Common limitations include arithmetic errors in complex calculations, misinterpretation of constraints, and failure to correctly apply specific mathematical properties or specialized factual knowledge, sometimes leading to confabulation.
