Gemini 3.5 Flash-Lite
Gemini 3.5 Flash-Lite is Google's low-latency, cost-effective multimodal model for high-throughput work: subagent tasks, document parsing, simple data extraction and high-volume agentic workflows.
Specifications & pricing.
- FIELD
- DOCUMENTED VALUE
- SOURCE
- Input modalities
Text, image, video, audio and PDF
Model page
- Output modality
Text
Model page
- Input token limit
1,048,576
Model page
- Output token limit
65,536
Model page
- Supported
Caching, code execution, file search, function calling, grounding with Google Maps, Search grounding, structured outputs, thinking, URL context
Model page
- Preview only
Computer use
Model page
- Not supported
Audio generation, image generation, Live API
Model page
- Consumption options
Batch API, Flex inference and Priority inference all supported
Model page
- Latest update
July 2026
Model page
- Standard price, paid tier
USD 0.30 input (text, image, video, audio) / USD 2.50 output per 1M tokens, output including thinking tokens, as published on 15 October 2026
- Context caching
USD 0.03 per 1M tokens plus hourly storage
Pricing page
- Batch and Flex
USD 0.15 input / USD 1.25 output per 1M tokens
Pricing page
- Priority
USD 0.54 input / USD 4.50 output per 1M tokens
Pricing page
When to use it
What Google says. The model page says the model is designed for "high-volume agentic workflows, simple data extraction, and applications where latency and API cost are the primary constraints". The pricing page adds translation and simple data processing, and the models list calls it Google's "fastest, most cost-effective 3.5 model for high-throughput execution". The same list tells new projects to use 3.5 Flash-Lite or 3.8 Flash.
Our editorial judgment. Flash-Lite looks like the worker tier in a routed system. A larger model plans; Flash-Lite handles the many small calls underneath: pulling fields from invoices, tagging support tickets, translating short strings, summarizing pages before a bigger model reads them. Where we would hesitate is open-ended reasoning or long multi-step coding, which Google aims at its larger Flash model. For consumer chat use, compare finished apps in our AI chatbots guide rather than raw API models.
Limits and costs to plan for
- Thinking tokens are billed as output. Google's output price includes thinking tokens. The model page lists thinking as supported but does not document levels, so test what your requests produce.
- Output is the bigger cost. Output is priced more than eight times input on Standard, so verbose answers and long thinking change your bill more than prompt length does.
- Text out only. Audio generation, image generation and the Live API are not supported.
- 65,536 output cap. Large generations may need chunking.
- No free-tier caching. Context caching is listed as not available on the free tier.
- Free tier data use. Google marks free tier traffic as used to improve its products; paid traffic is not.
- Grounding is metered. Search and Maps grounding include 5,000 free requests per month shared across Gemini 3 and newer models.
How it compares within the family
Gemini 3.5 Flash-Lite and Gemini 3.8 Flash share the same documented inputs, output, token limits and tool list. The documented differences are price, thinking detail, update date and positioning.
| Model | Standard input / output (USD per 1M tokens) | Thinking levels documented | Latest update |
|---|---|---|---|
| Gemini 3.5 Flash-Lite | 0.30 / 2.50 | "Supported", no levels listed | July 2026 |
| Gemini 3.8 Flash | 0.75 / 3.75 through 2026; 1.50 / 7.50 from 2027 | low, medium, high | September 2026 |
Flash-Lite has no scheduled price change listed, while 3.8 Flash doubles on 1 January 2027. Outside Google, DeepSeek's docs list DeepSeek V4.1 Flash at USD 0.15 input (cache miss) and USD 0.60 output per 1M tokens off-peak, and USD 0.30 and USD 1.20 at peak; DeepSeek V4 Pro sits higher at USD 1.32 and USD 3.96 peak.
How to evaluate it on your work
- Pick one narrow, high-volume task, such as field extraction or ticket tagging, and gather 100 or more labelled examples.
- Run them on Flash-Lite and on Gemini 3.8 Flash with the same prompt and a structured output schema.
- Score exact-match or schema-valid rates, and log input, output and thinking tokens per item.
- Price each run at Standard and at Batch or Flex rates, since many extraction jobs can wait.
- Route to Flash-Lite every task type where its accuracy meets your threshold, and send only the failures upward.
Method and limits
We read the Gemini 3.5 Flash-Lite model page, the Gemini models list and the Gemini API pricing page on 15 October 2026. We have not run Gemini 3.5 Flash-Lite, and nothing here is a test result or ranking. Prices and features change; check Google's pages before committing budget. See our ranking method for how HumanRank scores tools and the model directory for other profiles.
Frequently asked questions
What is Gemini 3.5 Flash-Lite? It is Google's low-latency, cost-effective multimodal model for high-throughput tasks such as subagent work, document parsing and simple data extraction.
How much does Gemini 3.5 Flash-Lite cost? It costs USD 0.30 per 1M input tokens and USD 2.50 per 1M output tokens on the paid Standard tier, with Batch and Flex at USD 0.15 and USD 1.25.
What is the Gemini 3.5 Flash-Lite context window? It is 1,048,576 input tokens, with a 65,536 token output limit.
Does Gemini 3.5 Flash-Lite support function calling? Yes, along with structured outputs, code execution, file search, URL context and both grounding options.
Can Gemini 3.5 Flash-Lite read images, audio and video? Yes, it accepts text, image, video, audio and PDF input, but it only outputs text.
Should I use Gemini 3.5 Flash-Lite or Gemini 3.8 Flash? Start with Flash-Lite for narrow, high-volume tasks and move to 3.8 Flash only where your tests show Flash-Lite falls short, since it is cheaper on both input and output.