Gemini 3.1 Flash Lite is now free on NetRoom

TL;DR

On May 7, 2026 Google shipped Gemini 3.1 Flash Lite. It's live on NetRoom from day one — free: no payment, no card, no VPN, no subscription. In short — a lightweight multimodal model with a 1,048,576-token context window and four thinking levels. Built for high-volume workloads: parsing, classification, lightweight agents.

What it is

Flash Lite sits at the bottom of the Gemini 3.1 Flash family. Google positions it for low latency and high request volume: high-throughput lightweight tasks where speed matters. Architecturally it's a multimodal transformer accepting text, images, video, audio and PDF inputs in a single request — no third-party converters.

Context is one million tokens. That's roughly 750 000 words or 2 000 pages. It fits a sizeable code repo, a full annual report, an hour-long transcript, or several hours of meeting notes. Recall through the first third of the window stays sharp on long inputs.

Thinking levels

Google added a thinking parameter to Flash Lite with four steps:

• minimal — near-zero reasoning, instant reply. For streaming workloads: classification, factual answers, template parsing.
• low — brief internal planning. Good for routine tasks where pace and accuracy both matter.
• medium — the default for multi-step questions.
• high — deep analysis: complex reasoning, multi-step problems, long analytical answers.

It's not a fast-vs-slow toggle but a thinking budget. Useful when an agent loop sits on top: minimal on routine steps, high at the hard branches.

What "free on NetRoom" actually means

No seven-day trials, no "10 messages a day" caps. Gemini 3.1 Flash Lite is available to every NetRoom user at no cost. You don't even need to sign up to try it — hit the homepage and start typing. Want history sync across devices and pinned chats? Sign up with email, no card.

Direct access to Gemini typically requires a VPN and a foreign-issued card. On NetRoom none of that applies: pick the model and go. Flash Lite stays free regardless — topping up on paid models doesn't change anything for Lite.

Where it fits

Four scenarios suit Flash Lite well:

• Parsing and extraction. Screenshot of a table, a PDF statement, an Excel document as an image — the model pulls structure without hand-crafted regex.
• Support and FAQ bots. A large context lets you load the whole knowledge base without a RAG pipeline for the prototype phase.
• Agent loops. Low latency plus native tool use — chains across dozens of steps stay snappy.
• Analytics. Reading a long report, a call transcript, a meeting log and pulling out conclusions — a job done in seconds.

When to reach for something heavier

Flash Lite doesn't replace the bigger models. Hard reasoning tasks — go to Gemini 3 Pro or Claude Sonnet. Native image generation — Nano Banana Pro and Flux are still the answer. Real-time web search — pick a model with search enabled. Flash Lite is a volume tool, optimised for the 90% of requests that are simple and need to fly.

How to try it

Open NetRoom, pick Gemini 3.1 Flash Lite from the model selector, send your first prompt. To compare against other models, flip the switch and send the same prompt to Gemini 3 Flash, GPT, or Claude. Full model card lives at /model/google/gemini-3.1-flash-lite/.