如果您正在尋找 Gemini 3 Pro API pricing, the most important update is that the original Gemini 3 Pro Preview API model is no longer the current model to build around.
According to Google’s Gemini API deprecation page, gemini-3-pro-預覽版 was shut down on March 9, 2026, and the recommended replacement is gemini-3.1-pro-preview.
That means old Gemini 3 Pro tutorials, pricing tables, or API examples may be outdated. If you are planning a new project, you should check the current Gemini model list and use the newer Gemini 3.1 Pro pricing as your main reference.
雙子星3號核心能力
雙子星3 Pro 概覽：
於 2025年11月18日, Gemini 3 Pro 是 Google 迄今為止最先進的人工智慧模型。它在以下領域表現卓越： 推理、多模態理解與程式碼生成, 使其適用於專業工作流程、複雜問題解決以及由人工智慧驅動的創意專案。.
- 深度推理 – Gemini 3 Pro 搭載創新的「深度思考」功能，能透過評估多重可能性並在回答複雜問題前自我邏輯驗證，實現類人深度思考能力。.
- 原生多模態 – 該模型從根本上設計為能無縫理解與處理文字、圖像、音訊、影片及程式碼。.
- 擴展上下文視窗 – 支援高達 100 萬個輸入語境單元與 64,000 個輸出單元，適用於處理長篇文件與複雜對話。.
- 代理式工作流程 – 專為自主編碼代理程式與複雜零次任務進行特別優化。透過新推出的 Bash 工具，它能導航檔案系統並驅動開發流程。.
- “「氛圍編碼」” – 支援從自然語言描述生成完整的應用程式，無需傳統程式設計語法。.
Gemini 3 Pro API 定價方案
Google最新旗艦模型Gemini 3 Pro已推出API預覽版，採用清晰且具競爭力的定價模式。相較於前代模型，Gemini 3 Pro根據上下文長度實施分級定價策略，旨在為大型專案與長文本處理優化成本。 對於低於20萬代幣的上下文，輸入/輸出價格分別為每百萬代幣$2.00/$12.00；超過20萬代幣時，價格將分別提升至$4.00與$18.00。.
核心定價概覽（每百萬代幣）
| 上下文長度 | 輸入價格 | 輸出價格 |
| ≤ 200,000 個代幣 | $2.00 | $12.00 |
| 200,000 枚代幣 | $4.00 | $18.00 |
Gemini 3 Pro API 定價機制詳解
Gemini 3 Pro API 定價圍繞兩個關鍵維度展開： 代幣數量 和 上下文視窗長度.
- 核心計費單位：代幣
與大多數大型語言模型（LLMs）類似，Gemini 3 Pro 對輸入代幣和輸出代幣分別計費，計費依據為 每百萬代幣（每100萬代幣）.
- 輸入 代幣（輸入價格）傳送至模型的資料量——您的提示、文件、程式碼等。.
- 輸出代幣（輸出價格）模型產生的回應、摘要、程式碼等內容的總量，包含用於「思考」的標記符。“
- 成本分級：情境視窗的重要性
Gemini 3 Pro 的突出特點在於其極長的上下文視窗（最多可達 一百萬個代幣為反映處理超長上下文的運算成本，Google 推出了分級定價方案：
| 上下文長度範圍 | 使用案例 | 定價費率 |
| 標準上下文（≤ 200K 詞元） | 日常閒聊、簡短摘要、標準任務 | 較低利率 |
| 長上下文（> 200K 詞元） | 法律文件分析、深度研究、複雜程式碼庫理解 | 較高利率 |
長上下文任務的輸入/輸出價格是標準上下文的兩倍，這反映出它們在複雜情境中的高價值。.
成本估算範例：開發人員如何編列預算
理解真實世界使用案例的成本有助於企業和開發人員精準規劃預算。以下是根據當前預覽版定價所估算的若干成本：
範例 1：標準客服聊天機器人（低情境）
假設：
- 輸入：5,000 個代幣
- 輸出：1,000 個代幣
- 上下文：5,000 個代幣（≤ 200K）
| 成本類型 | 計算 | 成本（美元） |
| 輸入 | 5,000 × $2.00 / 1M | 0.01 |
| 輸出 | 1,000 × $12.00 / 1M | 0.012 |
| 單次請求總數 | =$0.022 |
範例 2：文件分析與摘要（中長篇情境）
假設：
- 輸入：150,000 個標記（分析一份長篇報告）
- 輸出：8,000 個代幣（生成摘要）
- 上下文：150,000 個代幣（≤ 200K）
| 成本類型 | 計算 | 成本（美元） |
| 輸入 | 150,000 × $2.00 / 1M | 0.3 |
| 輸出 | 8,000 × $12.00 / 1M | 0.096 |
| 單次請求總數 | =$0.4 |
範例 3：深度研究任務（超長上下文）
假設：
- 輸入：350,000 個標記（分析多份文件）
- 輸出：15,000 個代幣（生成詳細報告）
- 上下文：350,000 個代幣（超過 200,000 個時，將適用更高費率）
| 成本類型 | 計算 | 成本（美元） |
| 輸入 | 350,000 × $4.00 / 1M | 1.4 |
| 輸出 | 15,000 × $18.00 / 1M | 0.27 |
| 單次請求總數 | =$1.67 |
Gemini 3 Pro vs Gemini 3.1 Pro Pricing
Gemini 3 Pro and Gemini 3.1 Pro are easy to confuse because the names are close, but they are not the same API pricing target anymore.
The older gemini-3-pro-預覽版 model has been shut down, so developers should not use it as the main pricing reference for new projects. The more useful comparison today is between old Gemini 3 Pro pricing expectations and the current Gemini 3.1 Pro pricing structure.
| 模型 | Current API status | 投入價格 | 輸出價格 | 這代表什麼 |
|---|---|---|---|---|
| Gemini 3 Pro Preview | Shut down | No longer current | No longer current | Do not base new API budgets on old gemini-3-pro-預覽版 examples. |
| Gemini 3.1 Pro Preview, prompts up to 200K tokens | Current replacement path | $2.00 / 1M 代幣 | $12.00 / 1M 代幣 | Main reference point for advanced Gemini API pricing. |
| Gemini 3.1 Pro Preview, prompts above 200K tokens | Current replacement path | $4.00 / 1M tokens | $18.00 / 1M tokens | Use this when working with very large context windows. |
| Gemini 3.1 Pro Preview, batch/flex up to 200K tokens | Lower-cost processing option | $1.00 / 1M 代幣 | $6.00 / 1M tokens | Useful for async workloads that do not need immediate standard-priority responses. |
The key point: Gemini 3 Pro is now mostly a legacy search term for API pricing, while Gemini 3.1 Pro is the better current pricing reference.
If you want the full cost breakdown, read our Gemini 3.1 Pro cost and pricing guide. If you are trying to understand the model differences, see our comparison of 雙子星 3.1 Pro vs 雙子星 3 Pro.
總結
Gemini 3 Pro pricing is no longer just a question of “how much does Gemini 3 Pro cost?” The more important question is which Gemini model you are actually using.
If you mean the old API model ID gemini-3-pro-預覽版, it has been shut down. For current developer projects, Gemini 3.1 Pro Preview is the safer pricing reference, especially for advanced reasoning, coding, long-context analysis, and multimodal workflows.
Before building or budgeting with Gemini API, check three things:
- Whether the model ID is still active.
- Whether your workload fits Pro, Flash, or Flash-Lite.
- Whether standard, batch, flex, or cached pricing gives the best value.
For most users, the practical path is simple: use Gemini 3.1 Pro for complex work, use lighter Gemini models for routine high-volume tasks, and compare alternatives like GPT, Claude, and Perplexity when quality, cost, or workflow flexibility matters. If you want to try multiple leading models in one place, GlobalGPT can help you compare Gemini-style workflows with other AI tools without switching between separate platforms.
常見問題
Is Gemini 3 Pro Preview still available in the API?
Google’s Gemini API deprecation page lists gemini-3-pro-預覽版 as shut down, with gemini-3.1-pro-preview as the recommended replacement. Developers should check the current model list before using old Gemini 3 Pro API examples.
What is the current Gemini 3.1 Pro API price?
On the paid standard tier, Gemini 3.1 Pro Preview starts at $2.00 per 1M input tokens and $12.00 per 1M output tokens for prompts up to 200K tokens. Prompts above 200K tokens cost more.
Is Gemini 3.1 Pro cheaper with batch or flex pricing?
Yes. Google’s Gemini API pricing lists lower batch and flex prices for Gemini 3.1 Pro Preview than the standard tier. These modes can be useful for workloads that do not need immediate standard-priority responses.
