Google Gemini 3.1 Pro first impressions: a 'Deep Think Mini' with adjustable reasoning on demand

Source: VentureBeat

Gemini 3.1 Pro – The First “Point‑One” Update to Google’s Frontier Model

For the past three months, Google’s Gemini 3 Pro has held its ground as one of the most capable frontier models available. But in the fast‑moving world of AI, three months is a lifetime — and competitors have not been standing still.

What’s New?

Earlier today Google released Gemini 3.1 Pro, an update that adds a three‑level adjustable‑thinking system (low / medium / high). This effectively turns the model into a lightweight version of Google’s specialized Deep Think reasoning system.

• First “point‑one” Gemini release – signals a shift from periodic full‑version launches to more frequent incremental upgrades.
• Three‑tier thinking gives developers a single model that can scale its reasoning effort dynamically:
  • Low – quick responses for routine queries.
  • Medium – balanced speed and depth (similar to the previous “high”).
  • High – “mini‑Deep Think” mode for multi‑minute, deep‑reasoning sessions.

The model is rolling out now in preview across:

• Gemini API via Google AI Studio
• Gemini CLI
• Google’s agentic development platform Antigravity
• Vertex AI
• Gemini Enterprise
• Android Studio
• Consumer Gemini app
• NotebookLM

The “Deep Think Mini” Effect: Adjustable Reasoning on Demand

The most consequential feature isn’t a single benchmark number; it’s the three‑tier thinking level system that gives fine‑grained control over computational effort per request.

Implications for enterprise deployment

• No longer need to route requests to different specialized models based on task complexity.
• Use a single model endpoint and adjust reasoning depth per task:
  • Routine document summarization → Low (fast).
  • Complex analytical tasks → High (Deep Think caliber).

Benchmark Performance – More Than Doubling Reasoning Over 3 Pro

Google’s published benchmarks show dramatic improvement, especially in reasoning and agentic capability.

The gains are especially relevant for production AI deployments that rely on tool use and multi‑step reasoning.

Why a “0.1” Release?

• Versioning shift – previous Gemini releases used dated previews (e.g., multiple 2.5 previews). Labeling this as 3.1 signals substantial improvements while framing it as an evolution, not a revolution.
• Technical basis – Google’s blog notes that 3.1 Pro builds directly on lessons from the Gemini Deep Think series, incorporating reinforcement‑learning (RL) techniques that boost performance on ARC‑AGI‑2, coding, and agentic benchmarks.
• Preview status – The model is still in preview; Google plans further advances in agentic workflows before moving to full General Availability (GA).

Competitive Implications for Your Enterprise AI Stack

For IT decision‑makers evaluating frontier models, Gemini 3.1 Pro forces a reassessment of both model selection and product‑development cadence.

• Benchmark leadership – 3.1 Pro now leads in several critical categories, pressuring Anthropic, OpenAI, and the open‑weight community to respond within weeks, not months.
• Operational simplicity – A single adjustable‑thinking model reduces the operational burden of maintaining multiple specialized endpoints.
• Future‑proofing – Rapid iteration cycles mean enterprises must adopt flexible integration strategies to keep pace with frequent model upgrades.

Availability

Gemini 3.1 Pro is available now in preview through the Gemini API in:

• Google AI Studio
• Gemini CLI
• Google Antigravity
• Vertex AI
• Gemini Enterprise
• Android Studio
• Gemini consumer app
• NotebookLM

Stay tuned for the GA rollout and further enhancements to the Deep Think reasoning stack.

Available Platforms

• Tigravity and Android Studio for developers.
• Enterprise customers can access it through Vertex AI and Gemini Enterprise.
• Consumers on Google AI Pro and Ultra plans can access it through the Gemini app and NotebookLM.