Gemini 3 Pro Preview Is Being Deprecated: How to Migrate to Gemini 3.1 Pro Before March 9

Google has announced that Gemini 3 Pro Preview will be fully retired on March 9, 2026. If you're using this model through the API — whether directly or via a proxy — you need to migrate to Gemini 3.1 Pro Preview before that date.

Here's what's changing, what you gain from the upgrade, and how to make the switch with minimal effort.

Timeline: What's Happening and When

Who Needs to Take Action?

You need to act if any of the following apply:

• You call gemini-3-pro-preview or gemini-pro-latest in your API requests
• You have pipelines, agents, or automation built on Gemini 3 Pro
• You use thinking_budget as a parameter (it's been replaced — more on that below)

Gemini 3.1 Pro vs 3 Pro: What's Actually Different?

This isn't just a version bump. Gemini 3.1 Pro Preview brings meaningful improvements, especially in reasoning.

Reasoning Performance

The headline number: Gemini 3.1 Pro scores 77.1% on ARC-AGI-2, more than double the score of Gemini 3 Pro. This translates to noticeably better performance on tasks that require multi-step logic, data synthesis, and complex code generation.

Input & Output

Features

Gemini 3.1 Pro supports the full feature set you'd expect from a frontier model:

• Thinking mode — built-in chain-of-thought reasoning
• Function calling — structured tool use
• Structured outputs — JSON schema enforcement
• Code execution — run code in a sandbox
• File search — query over uploaded documents

Breaking Change: thinking_budget → thinking_level

Migration Steps

The migration is straightforward — in most cases it's a one-line change.

Step 1: Update the Model String

Replace:

With:

Step 2: Update thinking_budget (If Used)

Step 3: Test Your Prompts

Gemini 3.1 Pro is generally better, but outputs may differ. Run your eval suite or spot-check critical prompts before going to production.

Code Example: Calling Gemini 3.1 Pro via EvoLink

curl -X POST "https://api.evolink.ai/v1beta/models/gemini-3.1-pro-preview:generateContent" \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "contents": [
      {
        "role": "user",
        "parts": [
          {
            "text": "Explain the key differences between TCP and UDP, with examples of when to use each."
          }
        ]
      }
    ]
  }'

curl -X POST "https://api.evolink.ai/v1beta/models/gemini-3.1-pro-preview:streamGenerateContent" \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "contents": [
      {
        "role": "user",
        "parts": [
          {
            "text": "Write a Python function that finds the longest palindromic substring in O(n) time."
          }
        ]
      }
    ]
  }'

Why Use EvoLink for Gemini 3.1 Pro?

Pricing

The pricing is identical for both Gemini 3 Pro and 3.1 Pro — so migrating doesn't cost you anything extra.

OpenAI SDK Compatibility

No need to learn a new SDK or rewrite your integration. If you're already using the OpenAI Python/Node SDK, you just point it at EvoLink's endpoint and swap the model name.

Both Models Available During Transition

TL;DR

• Gemini 3 Pro Preview dies March 9, 2026. The gemini-pro-latest alias switches to 3.1 Pro on March 6.
• Gemini 3.1 Pro is a significant upgrade — 2x+ reasoning performance, same feature set, same pricing.
• Migration is a one-line change in most cases: swap gemini-3-pro-preview → gemini-3.1-pro-preview.
• Watch for thinking_budget → thinking_level if you use the thinking feature.
• EvoLink supports both models now, with pricing up to 74% below Google's direct rates.