Gemini 3.5 Flash-Lite: Specs, Pricing & Key Features

Gemini 3.5 Flash-Lite is Google's fastest, most cost-effective model in the 3.5 class, launched on July 21, 2026. Designed for high-throughput, low-latency production workloads — from agentic search to document processing — it delivers 350 output tokens per second while costing just $0.30/1M input tokens. If you're evaluating which Gemini model fits your workflow, this guide breaks down what Gemini 3.5 Flash-Lite is, how it performs, how it compares to related models, and how you can access similar Gemini capabilities through Chat Smith today.

Gemini 3.5 Flash-Lite is a new member of Google DeepMind's Gemini 3.5 family, sitting below Gemini 3.5 Flash in terms of capability but well above older Flash-Lite variants in quality. It is optimized for scenarios where response speed and cost per call matter more than raw reasoning depth — think receipt extraction at scale, real-time agentic pipelines, and high-volume classification tasks.

According to Google, Gemini 3.5 Flash-Lite significantly outperforms Gemini 3.1 Flash-Lite across agentic and coding benchmarks. On SWE-Bench Pro, it scores 54.2% versus 31% for 3.1 Flash-Lite. On Terminal-Bench 2.1, the jump is even larger — 54% versus 31%. These are not marginal improvements; they represent a generational shift in what a "lite" model can do.

The model also supports flexible thinking levels — from minimal to high — letting developers trade off speed and cost against response quality depending on the task at hand. Computer use is now a built-in tool, making it suitable for automated web interactions and agentic task execution.

Gemini 3.5 Flash-Lite is the fastest model in the 3.5 series. At 350 output tokens per second as measured by Artificial Analysis, it is built for workloads where latency directly affects user experience or pipeline throughput. High-volume customer support, real-time classification, and streaming summarization are all natural fits.

At $0.30 per million input tokens and $2.50 per million output tokens, Gemini 3.5 Flash-Lite sits at a competitive price point — especially given the quality jump over 3.1 Flash-Lite. For teams processing millions of documents or running continuous agentic loops, this pricing can translate into significant infrastructure savings.

One of the standout features of 3.5 Flash-Lite is its support for multiple thinking levels: minimal, low, medium, and high. Minimal thinking prioritizes raw speed for latency-sensitive tasks. High thinking engages deeper reasoning for complex subagent workflows. This flexibility means a single model can serve multiple roles within a multi-agent architecture.

Like other Gemini models, 3.5 Flash-Lite accepts text, images, video, audio, and PDF inputs. This makes it practical for document understanding tasks — such as extracting line items from receipts or parsing structured data from scanned files — without needing to switch models for different input types.

Understanding where Gemini 3.5 Flash-Lite fits in the Gemini lineup helps you pick the right model for your use case. Here is how it stacks up against closely related variants:

Gemini 3.5 Flash-Lite is notably faster than Gemini 3.5 Flash while costing less per token — making it a strong choice when you need to run many parallel calls without a large model budget. On many agentic tasks, it even outperforms Gemini 3 Flash, which puts it in a surprisingly competitive position for a Lite-class model.

Gemini 3.5 Flash-Lite is already being used in production across several demanding verticals. Here are the strongest use cases:

Agentic search and retrieval. When a master agent like Gemini 3.6 Flash orchestrates a workflow, 3.5 Flash-Lite can handle sub-tasks in parallel — fetching data, categorizing results, and formatting output — at high speed and low cost.

Document processing. Ramp, a financial tech company, reported that 3.5 Flash-Lite landed on the Pareto frontier in their receipt extraction benchmark — offering an excellent balance of accuracy, latency, and cost for processing thousands of receipts daily.

Infrastructure data analysis. Ashler uses the model to analyze fragmented infrastructure records across hundreds of project deliverables, citing its ability to maintain context across complex, multi-step tasks.

Real-time content generation. Working alongside Gemini 3.6 Flash, 3.5 Flash-Lite has demonstrated the ability to instantly generate 25 unique web design concepts — showcasing how Lite-class models can power creative agentic applications when throughput matters.

Chat Smith does not currently include Gemini 3.5 Flash-Lite directly, but it gives you access to a broad set of Gemini models so you can find the right fit for your needs. You can use Gemini 3.5 Flash for complex reasoning and agents, or Gemini 2.5 Flash for a balance of speed and quality, or Gemini 2.5 Flash-Lite for maximum cost efficiency.

Chat Smith also gives you access to models from other leading labs — including GPT-5, Claude Sonnet 4.6, and Grok 4 — all from a single app. You can switch models mid-conversation, compare outputs side by side, and find the best model for each specific task without managing separate API keys or subscriptions.

Gemini 3.5 Flash-Lite represents a meaningful step forward in what a cost-efficient AI model can deliver. With 350 tokens per second, configurable thinking, multimodal input support, and significant improvements in agentic benchmarks over 3.1 Flash-Lite, it is a strong option for teams running high-volume production workloads. If you want to explore fast, capable Gemini models right now, Chat Smith gives you immediate access to the Gemini 3.5 Flash family alongside GPT-5, Claude, Grok, and more.

1. What is Gemini 3.5 Flash-Lite used for?

Gemini 3.5 Flash-Lite is designed for high-throughput, low-latency tasks such as agentic search, document processing, receipt extraction, real-time classification, and running sub-tasks inside larger multi-agent pipelines. Its configurable thinking levels make it flexible across both simple and moderately complex workflows.

2. How does Gemini 3.5 Flash-Lite differ from Gemini 3.5 Flash?

Gemini 3.5 Flash-Lite prioritizes speed and cost over raw capability. It runs at 350 tokens/second compared to roughly 250 for Gemini 3.5 Flash, and costs significantly less per token. For tasks that do not require deep multi-step reasoning, Flash-Lite delivers comparable quality at a fraction of the cost.

3. Is Gemini 3.5 Flash-Lite available in Chat Smith?

Gemini 3.5 Flash-Lite is not currently available directly in Chat Smith. However, Chat Smith offers several closely related Gemini models including Gemini 3.5 Flash and Gemini 2.5 Flash-Lite that cover similar use cases at comparable speed and cost.