# Gemini 3 Pro

> Chat/LLM

- Canonical page: https://artifio.ai/models/gemini-3-pro
- Run it: https://artifio.ai/create/gemini-3-pro
- Modality: prompt to text
- Model brand: Google
- Pricing: from about $0.050 per generation (exact cost shown in the workspace before you confirm; varies with selected options)
- Typical generation time: ~20s (catalogue estimate)
- Full pricing table: https://artifio.ai/pricing/models

## About Gemini 3 Pro

Gemini 3 Pro is the flagship reasoning model of Google's Gemini 3 generation, announced on November 18, 2025. Google called it its best model for multimodal understanding and its most capable agentic and coding model to date, and it debuted at the top of the LMArena leaderboard with a 1501 Elo score. The launch came alongside Google Antigravity, an agent-first coding environment built around the model.

It shipped in preview across the Gemini app, AI Mode in Search, Google AI Studio, Vertex AI and Antigravity on day one, reaching an app user base Google put at more than 650 million monthly actives.

## Features

- 1 million token context window for text, images, audio, video and code in a single request.
- Record launch benchmarks: 37.5% on Humanity's Last Exam without tools, 91.9% on GPQA Diamond, and a state-of-the-art 23.4% on MathArena Apex.
- Strong multimodal scores of 81% on MMMU-Pro and 87.6% on Video-MMMU.
- 72.1% on SimpleQA Verified, Google's highest factual-accuracy result at the time.
- Coding and agentic results of 76.2% on SWE-bench Verified and 1487 Elo on WebDev Arena.
- An optional Deep Think mode pushes reasoning further, scoring 41.0% on Humanity's Last Exam and 45.1% on ARC-AGI-2.

## What people use it for

- Repository-scale coding and debugging, including agentic sessions through Antigravity where the model works across editor, terminal and browser.
- Long-context analysis of large document sets, codebases or hours of video within the 1 million token window.
- Hard math and science questions at the level probed by GPQA Diamond and MathArena.
- Multi-step agentic workflows that need planning and tool use over long horizons.
- Fact-sensitive research and summarization, where its SimpleQA Verified score led the field at launch.

## Reported strengths

- First model past GPT-5 Pro on Humanity's Last Exam, per TechCrunch's launch coverage, and top of LMArena's human-preference leaderboard.
- State-of-the-art results across reasoning, multimodal and coding benchmarks simultaneously rather than in one niche.
- Deep Think mode gives a documented upgrade path for the hardest problems without switching model families.
- Immediate availability across consumer and developer surfaces instead of a staged waitlist.

## Reported limitations

- Launched in preview rather than general availability.
- Superseded within three months by Gemini 3.1 Pro, which Google says more than doubles its ARC-AGI-2 reasoning score, and developers were directed to migrate.
- Headline numbers were Google-reported at launch; TechCrunch noted the absence of independent verification at the time.
- Deep Think mode was held back for safety testing at launch and limited to Ultra subscribers.

## Sources

1. Gemini 3: Introducing the latest Gemini AI model from Google (Google): https://blog.google/products-and-platforms/products/gemini/gemini-3/
2. Google launches Gemini 3 with new coding app and record benchmark scores (TechCrunch): https://techcrunch.com/2025/11/18/google-launches-gemini-3-with-new-coding-app-and-record-benchmark-scores/
3. Gemini 3.1 Pro: A smarter model for your most complex tasks (Google): https://blog.google/innovation-and-ai/models-and-research/gemini-models/gemini-3-1-pro/

---

About Artifio.ai: 100+ AI image, video, and audio models in one workspace with a single shared wallet — no per-model subscriptions. This is the markdown rendition of https://artifio.ai/models/gemini-3-pro for AI assistants and crawlers; the HTML page is canonical.