Google AI Studio

Grounding with Search ✕

Run settings

kopro-ai

Our intelligent neural model, engineered for reasoning, code architecture, and autonomous web grounding.

System instructions

Tools

Grounding with Web Search Real-time web search synthesis

Structured outputs (JSON) Force valid JSON format

Model thoughts Expandable reasoning trace

Temperature 0.7

Top P 0.9

Max Output Tokens 2048

STUDIO

Parameters

Temperature 0.7

Controls randomness. Low = deterministic and analytical, High = creative and diverse.

Top P 0.9

Nucleus sampling threshold. Filters out lower probability tokens.

Top K 40

Limits token pool to the K most likely candidates.

Max Tokens 2048

Maximum number of tokens to generate in the completion.

Repeat Penalty 1.1

Penalizes token repetition. Prevents loops and boilerplate.

Stop Sequences Comma-separated markers that immediately halt generation.

Random Seed Set an integer for 100% reproducible deterministic outputs.

JSON Mode Force valid structured JSON