Gemini 3.1 Pro represents the newest flagship model from Google DeepMind, purpose-built for advanced reasoning, cross-domain synthesis, and high-level creative execution. It operates seamlessly across text, visual media, audio, video, and code, expanding upon the foundation of Gemini 3 Pro with stronger intelligence and more refined real-world comprehension.
Built for problems where simple answers fall short, 3.1 Pro applies advanced reasoning to interpret, synthesize, and deliver insights on intricate tasks from strategic data analysis to creative engineering workflows.
Gemini 3.1 Pro natively understands and processes text, imagery, audio, video, and code together. This lets it build richer outputs that combine these modalities seamlessly.
With the ability to handle large, multimodal inputs, 3.1 Pro excels at projects requiring deep context, such as system design, data synthesis, and long-form analysis.
Input:
Output (including thinking tokens):
From scientific reasoning to agentic coding and immersive 3D simulation, Gemini 3.1 Pro covers the full breadth of modern AI tasks with greater precision than any previous model in the Gemini family.
Gemini 3.1 Pro delivers smart, concise, and insightful answers on difficult academic and professional problems. Its reasoning prioritizes genuine intellectual depth over generic responses, scoring 44.4% on Humanity's Last Exam, the highest of any evaluated model without tool assistance.
The model natively processes text, images, video, audio, and PDF documents simultaneously. This enables tasks like analyzing a video alongside its transcript, extracting structured data from scanned PDFs, or reasoning over technical diagrams with supporting code.
Gemini 3.1 Pro leads several agentic coding benchmarks, including Terminal-Bench 2.0 (68.5%), SciCode (59%), and LiveCodeBench Pro Elo (2887). It brings best-in-class instruction following and tool use to complex, multi-file software development tasks, making it the top choice for "vibe coding" workflows.
Gemini 3.1 Pro achieves 92.6% on MMMLU, the leading multilingual knowledge benchmark — outperforming all other evaluated models. This makes it an exceptionally strong choice for global enterprise applications, cross-language research, and international product development.
Gemini 3.1 Pro was evaluated against the most recent competitive models across a comprehensive suite of academic and real-world benchmarks. Results below reflect Thinking (High) configuration unless otherwise noted.
No matter the scale of your ambition, whether technical, analytical, or creative, Gemini 3.1 Pro delivers the intelligence foundation required to build intelligent, adaptive, and future-ready AI workflows.
Build, debug, and refactor complex codebases using natural language. Its agentic coding capabilities make it ideal for ticket-to-PR workflows, infrastructure-as-code generation, and automated test writing across large mono-repos.
Accelerate literature synthesis, hypothesis testing, and data interpretation with a model trained to PhD-level scientific knowledge. Its 94.3% GPQA Diamond score demonstrates exceptional cross-disciplinary scientific reasoning.
Create intelligent agents and multi-tool pipelines using function calling, structured output, and the MCP standard. Gemini 3.1 Pro's industry-leading MCP Atlas score (69.2%) makes it the strongest backbone for sophisticated agent orchestration.
Measured on the AI/ML API gateway across live traffic. Figures are medians unless stated. Last updated 2026-10-12.
Gemini 3.1 Pro Preview is documented as suited to software engineering work and agentic workflows that need precise tool use and reliable multi-step execution. It is also suited to building with custom tools and bash, where it can prioritize those tools effectively. Another documented use case is creating code-based animated SVGs and other creative coding outputs directly from text prompts. A further stated use case is building complex interactive experiences, such as a live aerospace dashboard or sensory-rich 3D simulations.
import requests r = requests.post( "https://api.aimlapi.com/v1/chat/completions", headers={"Authorization": "Bearer " + AIMLAPI_KEY}, json={ "model": "google/gemini-3.1-pro-preview", "messages": [ { "role": "user", "content": "Hello!" } ] }, ) print(r.json())
const r = await fetch("https://api.aimlapi.com/v1/chat/completions", { method: "POST", headers: { Authorization: `Bearer ${process.env.AIMLAPI_KEY}`, "Content-Type": "application/json", }, body: JSON.stringify({ "model": "google/gemini-3.1-pro-preview", "messages": [ { "role": "user", "content": "Hello!" } ] }), }); console.log(await r.json());
curl -X POST https://api.aimlapi.com/v1/chat/completions \ -H "Authorization: Bearer $AIMLAPI_KEY" \ -H "Content-Type: application/json" \ -d '{"model":"google/gemini-3.1-pro-preview","messages":[{"role":"user","content":"Hello!"}]}'
OpenAI-compatible — swap the base URL and it works with your existing SDK.
| Type | Price |
|---|---|
| Input | |
| Output | |
| Cached input |
| Benchmark | Score | What it measures | Source | Retrieved |
|---|---|---|---|---|
| Humanity's Last Exam | 44.4% | Expert-level questions across many domains | Source | July 13, 2026 |
| Terminal-Bench | 68.5% | Autonomous shell/terminal task completion | Source | July 13, 2026 |
| SWE-bench Verified | 80.6% | Resolving verified real GitHub issues | Source | July 13, 2026 |
| GPQA Diamond | 94.3% | Google-proof graduate science questions (hardest subset) | Source | July 13, 2026 |
| Intelligence | 30.4 | Composite score across standardised reasoning, knowledge and problem-solving evaluations, measured independently by Artificial Analysis | Source | September 12, 2026 |
| Coding | 68.8 | Composite score across standardised coding evaluations, measured independently by Artificial Analysis | Source | September 12, 2026 |
| Model | Input | Output | Context | Best for |
|---|---|---|---|---|
Gemini 3.1 Pro This page | Coding + agents | |||
| Reasoning + agents | ||||
| Balanced coding + agents | ||||
| Long-context, multimodal & agentic workflows | ||||
| Reasoning + agents |
Gemini 3.1 Pro has a 1,000,000 tokens context window and can return up to 65,536 tokens.
Gemini 3.1 Pro takes image, text as input and returns text.
Use google/gemini-3.1-pro-preview as the model id. Requests go to https://api.aimlapi.com/v1/chat/completions.
Gemini 3.1 Pro became available on February 25, 2026.
Gemini 3.1 Pro is priced at input $2.6 / 1M tokens, output $15.6 / 1M tokens, cached input $0.65 / 1M tokens.
Yes, Gemini 3.1 Pro can stream responses as they are generated.
Yes, Gemini 3.1 Pro accepts image input alongside text.
Gemini 3.1 Pro was built by Google.
It is described as a reasoning model optimized for software engineering and agentic workflows.
Yes, it supports function calling, tool use, and parallel tool calls, along with structured output.
