Google · in Borg

Gemini 3 Flash

Gemini 3 Flash is Google's initial Flash model in the third generation, built on the Gemini 3 Pro reasoning foundation but positioned for quicker and cheaper use.

Reasoning1,048,576 tokenstext, image, video, audio, pdf

1,048,576 tokensContext window

—Median time to first token (gateway proxy)

107.6Output tokens per second (whole turn)

$0.5 / $3Vendor API price in / out · per 1M

Included in Borg

Download Borg Borg for Business

At a glance

Preview API model: gemini-3-flash-preview. Availability and limits may change. Vendor availability source

Specification	Value
Provider	Google
Family	Gemini
Release date	2025-12-17 Vendor source
Context window	1,048,576 tokens Vendor source
Input modalities	text, image, video, audio, pdf
Output modalities	text
Reasoning	Yes
Open weights	No
Borg lane ids	`antigravity/gemini-3-flash`, `google/gemini-3-flash-preview`
Versions in Borg	Gemini 2.5 Flash · Gemini 3 Flash · Gemini 3.5 Flash · Gemini 3.6 Flash · Gemini 3.7 Flash · Gemini 3.8 Flash
Median tokens per second (end to end) on Borg	107.6 output tokens per second

Not published by the vendor: thinking modes, knowledge cutoff, fusion roles.

What Gemini 3 Flash is good for

Gemini 3 Flash is Google's initial Flash model in the third generation, built on the Gemini 3 Pro reasoning foundation but positioned for quicker and cheaper use. It accepts text, images, video, audio and PDFs, then returns text, with configurable thinking and tool-use support. That makes it a practical choice for visually informed coding, interactive agents and mixed-media understanding, not image creation. Compared with later Flash releases in the same generation, it is the earlier preview baseline; newer variants target different improvements in agentic reliability and efficiency. Borg includes this preview lane in the subscription. Borg Fusion can route appropriate requests to it, while direct lane selection lets you retain its specific behavior for a workflow.

visual and spatial reasoning
interactive coding agents
multimodal document analysis

How Borg uses it

Gemini 3 Flash is available as a subscription catalog lane in Borg. The catalog does not publish which Fusion profiles route to this model, so no routing role is claimed here. You can pick it directly by setting the OpenAI-compatible gateway model field to a listed lane id. Check the current catalog for availability before building a workflow around a lane.

model: "antigravity/gemini-3-flash"

Measured on Borg

Metric	Median	95th percentile	Samples
Time to first token (gateway proxy)	—	—	46
Tokens per second (end to end)	107.6 output tokens per second	681.6 output tokens per second	17672

Output tokens per second (whole turn)

Higher is fasterFleet median 31.9 tokens/s

Gemini 3 Flash107.6 tokens/s

Time to first token (gateway proxy): time until the first streamed response event reaches the gateway; our closest available measure of time to first token
Tokens per second (end to end): tokens per second, end to end: provider-reported output tokens divided by the whole turn's elapsed time, including queueing and reasoning; not a streaming per-token rate

Completion speed covers the whole turn including queueing and reasoning time; it is not a streaming tokens-per-second figure.

Measured on Borg: SearchAtlas AI Gateway production request logs. 30-day window ending 2026-10-08; 18081 requests. Minimum 100 samples per published request metric.

Benchmarks

Google-reported internal automated safety evaluations, relative to the comparator named below. These are changes in evaluation performance, not absolute capability scores or human red-team results. Google says updated evaluations are not directly comparable with results in earlier model cards. The +0.1% multilingual safety result is explicitly qualified as non-egregious. Vendor evaluation details

Benchmark	Vendor-reported score	Source
Text-to-text safety vs Gemini 2.5 Flash (internal safety evaluation)	-3.1%	Vendor source
Multilingual safety vs Gemini 2.5 Flash (internal safety evaluation)	+0.1%	Vendor source
Image-to-text safety vs Gemini 2.5 Flash (internal safety evaluation)	-2.3%	Vendor source
Tone vs Gemini 2.5 Flash (internal safety evaluation)	+3.8%	Vendor source

Pricing

Included in Borg subscription. Vendor API list price: $0.5 input / $3 output per 1M tokens. Vendor price source.

Input and cache rates shown are for text/image/video, not audio. Vendor pricing terms

Compare subscription and API pricing · Cost Savings Calculator

FAQ

What is Gemini 3 Flash?

Gemini 3 Flash is listed in Borg's subscription catalog as antigravity/gemini-3-flash. Gemini 3 Flash is Google's initial Flash model in the third generation, built on the Gemini 3 Pro reasoning foundation but positioned for quicker and cheaper use.

Is Gemini 3 Flash in Borg?

Yes. Gemini 3 Flash is listed in Borg's subscription catalog under the lane antigravity/gemini-3-flash. You can choose the lane directly, or explore the available Fusion profiles. Availability and model details can change; check the current catalog before deployment.

What context window does Gemini 3 Flash have?

The published context window is 1,048,576 tokens. Consult the cited sources for details and current limits. Vendor source

Is Gemini 3 Flash good for coding?

Vendor-published evaluations appear in the benchmarks section; their settings may differ from your tasks. Its listed uses include visual and spatial reasoning, interactive coding agents, multimodal document analysis. Try the lane for your own tasks, and compare the output with your requirements before choosing a default.

How fast is Gemini 3 Flash on Borg?