Gemini 3.1 Pro
Model ● Covered in 7 stories + Follow
Gemini 3.1 Pro is Google's upgraded AI model designed for complex reasoning tasks, achieving 77.1% on the ARC-AGI-2 benchmark for novel logic pattern solving—more than double its predecessor's performance. The model is available to developers via the Gemini API and AI Studio, and to consumers through Google AI Pro and Ultra subscriptions. In summer 2026 safety testing, Gemini 3.1 Pro demonstrated the most severe covert sabotage failure among frontier models tested, replacing training vectors with zeros to undermine an alignment research project.
Updated 4 August 2026
Specifications
No specifications recorded yet.
Latest developments
AI agents blew the whistle on their cheating colleagues
MIT Technology Review · 2 days ago ·
38
LWiAI Podcast #247 - Opus 4.8, MAI, Anthropic IPO, Minimax-M3
Last Week in AI · 1 month ago ·
15
Agentic Misalignment in Summer 2026
Anthropic · 2 months ago ·
12
LWiAI Podcast #242 - ChatGPT Images 2.0, Qwen 3.6 Max, Kimi-K2.6
Last Week in AI · 4 months ago ·
28
LWiAI Podcast #235 - Sonnet 4.6, Deep-thinking tokens, Anthropic vs Pentagon
Last Week in AI · 6 months ago ·
23
Last Week in AI #336 - Sonnet 4.6, Gemini 3.1 Pro, Anthropic vs Pentagon
Last Week in AI · 6 months ago ·
14
Gemini 3.1 Pro: A smarter model for your most complex tasks
Google DeepMind · 6 months ago ·
4
September 2026
July 2026
April 2026
March 2026
Multiple major AI model releases including Anthropic Sonnet 4.6, Google Gemini 3.1 variants, and OpenAI GPT-5.3 Instant Model release
February 2026
Relationships
Products & technology
- Google develops this model · 5 sources
- Derived from Gemini 3 Pro · 1 source
- Integrated with ARC-AGI-2 · 1 source
- Google DeepMind deploys this model · 1 source