Gemini 3.1 Flash-Lite is Google's cheap, fast API model for high-volume work: sorting emails, pulling fields out of PDFs, translating short texts. It costs $0.25 per million input tokens. For new projects Google now recommends Gemini 3.5 Flash-Lite, and 3.1 may be retired from May 2027. Use the paid tier for anything real.
Other good fits
Reading long inputs cheaply: 1 million tokens of context, with text, images, audio, video and PDF as input.
Prototyping in Google AI Studio for free before wiring the model into an automation.
+ Pros
Read pros
Flash-Lite is the budget line of Google's Gemini models, sold through the Gemini API. The catalogue entry names Gemini 3.1 Flash-Lite (model ID gemini-3.1-flash-lite, stable since May 2026). Google has since released Gemini 3.5 Flash-Lite (gemini-3.5-flash-lite, July 2026) and recommends it for new projects. Both take text, images, audio, video and PDFs, and both read up to about a million tokens at once.
The point of Flash-Lite is volume. When one call costs a fraction of a cent, you can run the same instruction over every row of a spreadsheet or every PDF in a folder.
Pick it when:
You have a repetitive, simple job: classify each incoming email, pull five fields from each delivery note, translate short product texts.
You need structured output (JSON with fixed fields) that a spreadsheet or automation can read.
You'd run the job overnight: the Batch API halves the price.
Small models make small mistakes confidently. Misread digits, swapped dates, a guessed value when the field is missing. Always compare a sample against the source.
Cloud only. Your documents go to Google. Fine for public or low-risk data on the paid tier; check with your data protection officer before sending personal data.
Model versions move quickly. Google releases new Flash models every few months and retires older ones on a published schedule.
Data may not leave your building: run Gemma 4 on your own machine with Ollama.
You want one bill across many model vendors: OpenRouter.
Avoid if
Data must stay on your own hardware. Run an open model such as Gemma 4 locally.
Context
Use the live model list and API response metadata to confirm the exact Gemini 3.1 Flash-Lite identifier, modalities and limits for the account and endpoint in use.
Requirements
A Google AI Studio or Vertex AI route, project credentials and a supported API client are required.
Licence & cost
Read licence & cost
Prices checked 23 September 2026 (Gemini API, USD per million tokens, standard paid tier):
Model
Input
Output
Batch (input / output)
Gemini 3.1 Flash-Lite
$0.25 (text, image, video); $0.50 (audio)
$1.50
$0.125 / $0.75
Gemini 3.5 Flash-Lite
$0.30
$2.50
$0.15 / $1.25
Both have a free tier with lower rate limits. Outside Europe, Google uses free-tier content to improve its products; on the paid tier it doesn't. In the EEA, Switzerland and the UK, Google's terms apply the paid-tier data rules to free use too, but any app you offer to users there must run on the paid tier.
At these prices, 1,000 one-page invoices typically cost well under a dollar. The effort is in the setup: an API key kept out of shared files, a budget alert on the billing account, and a spot check of results every time you change the prompt or the model.
Plan the model switch now. Google lists the earliest shutdown for gemini-3.1-flash-lite as 7 May 2027, with 3.5 Flash-Lite as the replacement. Write the model ID in one place so changing it is one edit.
First task
Documentation-based suggestion — not run here.
Read first task
Extract fields from 20 invoices into a table. Allow 45 minutes; you need a Google account and, for real documents, a paid API key.
Make or collect 20 sample invoices as PDFs. Use invented suppliers and amounts for the first run.
In Google AI Studio, choose gemini-3.5-flash-lite (or gemini-3.1-flash-lite), switch on structured output, and define five fields: supplier, invoice number, date, net amount, VAT amount.
Upload one PDF with the instruction: "Extract these fields. If a field is missing, return null. Never guess." Check the result against the PDF.
Run the other 19 and paste the results into a sheet next to the correct values.
Success check: at least 19 of 20 rows match exactly, and every missing field came back as null rather than an invented value. If it misses more, give it two worked examples in the prompt and try again. To run this on a schedule, put the call inside an automation; see T05-L02.
Links & evidence
Documentation-based · not hands-on · Sep 23, 2026
Checked against Google's Gemini API model list, model pages, pricing page, deprecation schedule and API terms on 23 September 2026. We did not send API requests for this review.
Gemini API pricing · Sep 23, 2026 — 3.1 Flash-Lite $0.25 input (text/image/video), $0.50 audio, $1.50 output per 1M tokens; 3.5 Flash-Lite $0.30 / $2.50; Batch halves the price; free tier content used to improve products.
Gemini deprecations · Sep 23, 2026 — gemini-3.1-flash-lite released 7 May 2026, earliest shutdown 7 May 2027, replacement gemini-3.5-flash-lite.
Gemini models · Sep 23, 2026 — 3.1 Flash-Lite and 3.5 Flash-Lite are stable; Google recommends 3.5 Flash-Lite or 3.8 Flash for new projects.
Gemini 3.1 Flash-Lite model page · Sep 23, 2026 — 1,048,576 input tokens, 65,536 output tokens; text, image, video, audio and PDF input; structured outputs supported.
Gemini API Additional Terms · Sep 23, 2026 — In the EEA, Switzerland and the UK the paid-service data terms apply to all use, and apps offered to users there must use paid services.
Further review notes & alternatives
Alternatives and trade-offs
GPT mini models win when your team already has an OpenAI account and wants one vendor for everything.
Gemma 4 wins when privacy beats convenience: open weights, runs locally, no per-token cost, but you need the hardware.
Gemini 3.1 Pro wins when quality matters more than price per item, at roughly eight times the cost.
Process text, images, video, audio, and PDFs in bulk.
Use its large context for document or transcript extraction.
Choose Vertex AI regions when an available EU location fits.
