google/gemini-3-flash

Google's most intelligent model built for speed with frontier intelligence, superior search, and grounding

Capabilities

Top-P

Cost

Community model (estimated from hardware time)

Input Parameters

Name	Type	Description	Default	Constraints
`prompt`*	string	The text prompt to send to the model	`—`	—
`audio`	string(uri)	Input audio to send with the prompt (max 1 audio file, up to 8.4 hours)	`—`	—
`images`	array	Input images to send with the prompt (max 10 images, each up to 7MB)		—
`max_output_tokens`	integer	Maximum number of tokens to generate	`65535`	min: 1, max: 65535
`system_instruction`	string	System instruction to guide the model's behavior	`—`	—
`temperature`	number	Sampling temperature between 0 and 2	`1`	min: 0, max: 2
`thinking_level`	string	Thinking level for reasoning (low or high). Replaces thinking_budget for Gemini 3 models.	`"none"`	nonelowhigh
`top_p`	number	Nucleus sampling parameter - the model considers the results of the tokens with top_p probability mass	`0.95`	min: 0, max: 1
`video_fps`	number	Frame rate (frames per second) at which to sample input videos. Higher values give the model more frames to reason over (useful for fast-moving footage) but consume more tokens. Omit to use the default sampling rate (1 fps).	`—`	min: 0.1, max: 60
`videos`	array	Input videos to send with the prompt (max 10 videos, each up to 45 minutes)		—

promptrequiredstring

The text prompt to send to the model

audiostring

Input audio to send with the prompt (max 1 audio file, up to 8.4 hours)

imagesarray

Input images to send with the prompt (max 10 images, each up to 7MB)

Default:

max_output_tokensinteger

Maximum number of tokens to generate

Default: 65535min: 1, max: 65535

system_instructionstring

System instruction to guide the model's behavior

temperaturenumber

Sampling temperature between 0 and 2

Default: 1min: 0, max: 2

thinking_levelstring

Thinking level for reasoning (low or high). Replaces thinking_budget for Gemini 3 models.

Default: "none"

nonelowhigh

top_pnumber

Nucleus sampling parameter - the model considers the results of the tokens with top_p probability mass

Default: 0.95min: 0, max: 1

video_fpsnumber

Frame rate (frames per second) at which to sample input videos. Higher values give the model more frames to reason over (useful for fast-moving footage) but consume more tokens. Omit to use the default sampling rate (1 fps).

min: 0.1, max: 60

videosarray

Input videos to send with the prompt (max 10 videos, each up to 45 minutes)

Default:

Version: a4cab3326ec3Updated: 7/25/20265.1M runs