Google
Language model

Gemini 3.1 Pro Preview

Google's refreshed frontier model with sharper reasoning.

4.8

At a glance

Provider
Google
Type
Language model
Context window
1,048,576 tokens
Released
Early 2026
Model ID
google/gemini-3.1-pro-preview

Overview

Gemini 3.1 Pro is Google DeepMind's first major refresh of the Gemini 3 Pro line, pushing its already-strong reasoning and multimodal scores higher. It keeps the 1-million-token context window and 64K output window of its predecessor and tightens performance on GPQA, AIME and LiveCodeBench — the benchmark suites where Gemini 3 first made its name.

Its superpower remains multimodal reasoning over huge inputs: long documents, whole codebases, images and video analysed together at a depth rivals struggle to match. With Gemini 3.5 Pro rumoured next, 3.1 Pro is the current Pro-tier default for the most demanding long-context and multi-modal work.

What it can do

  • 1M-token context, 64K-token output
  • Improved reasoning over Gemini 3 Pro
  • Top-tier multimodal and coding benchmarks
  • Strong on GPQA, AIME and LiveCodeBench

Pricing: Premium frontier tier; cheaper Flash variant available.

Benchmarks

Context
1M tokens
Output
64K tokens
GPQA
~94%

Pros and cons

What works

  • Massive 1M-token context window
  • Best-in-class multimodal reasoning
  • Genuine gains over Gemini 3 Pro

Trade-offs

  • Premium Pro-tier pricing
  • Quality can be uneven outside its strongest categories
  • Gemini 3.5 Pro expected soon

The verdict

Gemini 3.1 Pro is the refined frontier pick for long-context and multimodal reasoning — analysing whole codebases, documents or video in one pass. It tightens Gemini 3 Pro's strengths and is the model to beat at huge context lengths, even as everyone waits for 3.5 Pro.

Best for: Long-context and multimodal reasoning at the frontier.

Questions

How is Gemini 3.1 Pro different from 3 Pro?

It's a refresh of Gemini 3 Pro with improved reasoning and multimodal scores — better on GPQA, AIME and LiveCodeBench — while keeping the same 1M context and 64K output windows.

Is Gemini 3.1 Pro better than GPT-5.5?

They trade blows. Gemini 3.1 Pro leads on long-context and multimodal reasoning; GPT-5.5 is prized for agentic tool use and reliability. Comparing them side-by-side on your own prompts is the best way to decide.

Further reading

Put Gemini 3.1 Pro Preview at the table.

Ask Gemini 3.1 Pro Preview and other models the same question side by side, then keep the best answer.