Key takeaways

• Gemini 3.7 Flash is a Transformer-based, multimodal model designed for fast agentic workflows, coding, and complex multi-step reasoning, accepting various input types including text, images, audio, video, and files.
• The model demonstrates exceptional mathematical problem-solving capabilities, achieving 98.82% accuracy on the MATH-500 benchmark, excelling in complex operations, logical deductions, and applying mathematical theorems.
• A significant limitation was observed in the Multimodal Understanding benchmark, where the model failed to generate any responses for 900 prompts, resulting in 0% accuracy and indicating a critical flaw in processing multimodal multiple-choice questions.
• While strong in mathematical reasoning, minor issues were identified in strict adherence to specific output formatting (e.g., unit notation, interval notation prefixes) rather than mathematical correctness.
• The model's strong performance on mathematical tasks suggests its potential for applications requiring precise mathematical reasoning and computation, despite the critical failure in multimodal understanding.

Technical specifications

Context length

1M tokens

Max output

—

Architecture

Transformer

Modalities

text+image+file+audio+video->text

Parameters

—

Open weights

No

Licence

Proprietary

Regions

USA

Vendor

Google

Provider model key

google/gemini-3.7-flash