Gemini Deep Research 現已推出預先發布版，提供協作規劃、視覺化、MCP 支援等功能。

Google uses AI technology to translate content into your preferred language. AI translations can contain errors.

Interactions API：重大變更遷移指南 (2026 年 5 月)

v1beta Interactions API 導入破壞性變更，重新架構 API 形狀，以支援飛行中導引和非同步工具呼叫等未來能力。本頁說明異動內容，並提供異動前後的程式碼範例，協助您完成遷移。變更分為兩類：

步驟結構定義：新的 steps 陣列會取代 outputs 陣列，提供每個互動回合的結構化時間軸。
輸出格式設定：新的多型 response_format 會整合所有輸出格式控制項，並移除 response_mime_type。

請按照「如何遷移至新版結構定義」一文中的步驟，更新整合服務。

核心異動：`outputs` 改為 `steps`

新結構定義會將 outputs 陣列替換為 steps 陣列。

舊版：傳回的回覆是扁平的 outputs 陣列，只包含模型生成的內容。
新結構定義：回覆會傳回 steps 陣列，其中包含具有型別鑑別器的結構化步驟。

POST /interactions 僅會傳回輸出步驟。GET /interactions/{id} 會傳回完整步驟時間軸，包括初始 user_input 步驟。

基本輸入/輸出 (一元)

之前 (舊版)

Python

# Request
interaction = client.interactions.create(
    model="gemini-3-flash-preview", input="Tell me a joke."
)

# Response access
print(interaction.outputs[-1].text)

JavaScript

// Request
const interaction = await client.interactions.create({
    model: 'gemini-3-flash-preview',
    input: 'Tell me a joke.'
});

// Response access
console.log(interaction.outputs[-1].text);

REST

curl -X POST "https://generativelanguage.googleapis.com/v1beta/interactions?key=$GEMINI_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gemini-3-flash-preview",
    "input": "Tell me a joke."
  }'

// Response
{
  "id": "int_123",
  "role": "model",
  "outputs": [
    {
      "type": "text",
      "text": "Why did the chicken cross the road?"
    }
  ]
}

之後 (新結構定義)

Python

# Request
interaction = client.interactions.create(
    model="gemini-3-flash-preview", input="Tell me a joke."
)

# Response access
print(interaction.steps[-1].content[0].text)  # CHANGED: steps instead of outputs

JavaScript

// Request
const interaction = await client.interactions.create({
    model: 'gemini-3-flash-preview',
    input: 'Tell me a joke.'
});

// Response access
console.log(interaction.steps.at(-1).content[0].text);

REST

# Opt-in needed before May 26th
curl -X POST "https://generativelanguage.googleapis.com/v1beta/interactions?key=$GEMINI_API_KEY" \
  -H "Content-Type: application/json" \
  -H "Api-Revision: 2026-05-20" \
  -d '{
    "model": "gemini-3-flash-preview",
    "input": "Tell me a joke."
  }'

// POST Response
{
  "id": "int_123",
  "steps": [
    {
      "type": "model_output",
      "content": [
        {
          "type": "text",
          "text": "Why did the chicken cross the road?"
        }
      ]
    }
  ]
}

// GET /v1beta/interactions/int_123 (returns full timeline including input)
{
  "id": "int_123",
  "steps": [
    {
      "type": "user_input",
      "content": [
        { "type": "text", "text": "Tell me a joke." }
      ]
    },
    {
      "type": "model_output",
      "content": [
        {
          "type": "text",
          "text": "Why did the chicken cross the road?"
        }
      ]
    }
  ]
}

函式呼叫

要求結構維持不變，但回應會以結構化步驟取代平面 outputs 內容。

之前 (舊版)

Python

# Accessing function call in legacy schema
for output in interaction.outputs:
    if output.type == "function_call":
        print(f"Calling {output.name} with {output.arguments}")

JavaScript

// Accessing function call in legacy schema
for (const output of interaction.outputs) {
    if (output.type === 'function_call') {
        console.log(`Calling {output.name} with {JSON.stringify(output.arguments)}`);
    }
}

REST

// Response
{
  "id": "int_001",
  "role": "model",
  "status": "requires_action",
  "outputs": [
    {
      "type": "thought",
      "signature": "abc123..."
    },
    {
      "type": "function_call",
      "id": "fc_1",
      "name": "get_weather",
      "arguments": { "location": "Boston, MA" }
    }
  ]
}

之後 (新結構定義)

Python

# Accessing function call in new steps schema
for step in interaction.steps:
    if step.type == "function_call":
        print(f"Calling {step.name} with {step.arguments}")

JavaScript

// Accessing function call in new steps schema
for (const step of interaction.steps) {
    if (step.type === 'function_call') {
        console.log(`Calling {step.name} with {JSON.stringify(step.arguments)}`);
    }
}

REST

// POST Response
{
  "id": "int_001",
  "status": "requires_action",
  "steps": [
    {
      "type": "thought",
      "summary": [{
        "type": "text",
        "text": "I need to check the weather in Boston..."
      }],
      "signature": "abc123..."
    },
    {
      "type": "function_call",
      "id": "fc_1",
      "name": "get_weather",
      "arguments": { "location": "Boston, MA" }
    }
  ]
}

伺服器端工具

伺服器端工具 (例如 Google 搜尋或程式碼執行) 現在會在 steps 陣列中產生特定步驟類型。舊版架構會在 outputs 陣列中將這些作業傳回為特定內容類型，新版架構則會將這些作業移至 steps 陣列。下列範例使用 Google 搜尋。

之前 (舊版)

Python

# Accessing search results in legacy schema
for output in interaction.outputs:
    if output.type == "google_search_call":
        print(f"Searched for: {output.arguments.queries}")
    elif output.type == "google_search_result":
        print(f"Found results: {output.result.rendered_content}")

JavaScript

// Accessing search results in legacy schema
for (const output of interaction.outputs) {
    if (output.type === 'google_search_call') {
        console.log(`Searched for: {output.arguments.queries}`);
    } else if (output.type === 'google_search_result') {
        console.log(`Found results: {output.result.renderedContent}`);
    }
}

REST

curl -X POST "https://generativelanguage.googleapis.com/v1beta/interactions?key=$GEMINI_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gemini-3-flash-preview",
    "input": "Who won the last Super Bowl?",
    "tools": [
      { "type": "google_search" }
    ]
  }'

// Response
{
  "id": "int_456",
  "outputs": [
    {
      "type": "google_search_call",
      "id": "gs_1",
      "arguments": { "queries": ["last Super Bowl winner"] }
    },
    {
      "type": "google_search_result",
      "call_id": "gs_1",
      "result": {
        "rendered_content": "<div>...</div>",
        "url": "https://www.nfl.com/super-bowl"
      }
    },
    {
      "type": "text",
      "text": "The Kansas City Chiefs won the last Super Bowl.",
      "annotations": [
        {
          "start_index": 4,
          "end_index": 22,
          "source": "https://www.nfl.com/super-bowl"
        }
      ]
    }
  ],
  "status": "completed"
}

之後 (新結構定義)

Python

# Accessing search results in new steps schema
for step in interaction.steps:
    if step.type == "google_search_call":
        print(f"Searched for: {step.arguments.queries}")
    elif step.type == "google_search_result":
        print(f"Found results: {step.result.search_suggestions}")

JavaScript

// Accessing search results in new steps schema
for (const step of interaction.steps) {
    if (step.type === 'google_search_call') {
        console.log(`Searched for: {step.arguments.queries}`);
    } else if (step.type === 'google_search_result') {
        console.log(`Found results: {step.result.searchSuggestions}`);
    }
}

REST

# Opt-in needed before May 26th
curl -X POST "https://generativelanguage.googleapis.com/v1beta/interactions?key=$GEMINI_API_KEY" \
  -H "Content-Type: application/json" \
  -H "Api-Revision: 2026-05-20" \
  -d '{
    "model": "gemini-3-flash-preview",
    "input": "Who won the last Super Bowl?",
    "tools": [
      { "type": "google_search" }
    ]
  }'

// POST Response
{
  "id": "int_456",
  "steps": [
    {
      "type": "google_search_call",
      "id": "gs_1",
      "arguments": { "queries": ["last Super Bowl winner"] },
      "signature": "abc123..."
    },
    {
      "type": "google_search_result",
      "call_id": "gs_1",
      "result": {
        "search_suggestions": "<div>...</div>"
      },
      "signature": "abc123..."
    },
    {
      "type": "model_output",
      "content": [
        {
          "type": "text",
          "text": "The Kansas City Chiefs won the last Super Bowl.",
          "annotations": [
            {
              "type": "url_citation",
              "url": "https://www.nfl.com/super-bowl",
              "title": "NFL.com",
              "start_index": 4,
              "end_index": 22
            }
          ]
        }
      ]
    }
  ],
  "status": "completed"
}

串流

串流會公開新的事件類型：

新事件類型

interaction.created
interaction.completed
interaction.in_progress
interaction.requires_action
interaction.error
step.start
step.delta
step.stop

已淘汰的事件類型

上述新事件會取代下列舊版事件類型：

interaction.start → interaction.created
content.start → step.start
content.delta → step.delta
content.stop → step.stop
interaction.complete → interaction.completed
error → interaction.error
interaction.status_update → 取代為 interaction.in_progress、interaction.requires_action 等。

串流函式呼叫：使用串流函式呼叫時，step.start 事件會傳送函式名稱，而 step.delta 事件會將引數串流為部分 JSON 字串 (使用 arguments_delta)。您必須累計這些差異，才能取得完整引數。這與一元呼叫不同，因為您會一次收到完整的函式呼叫物件。

範例

之前 (舊版)

Python

# Legacy streaming used content.delta
stream = client.interactions.create(
    model="gemini-3-flash-preview",
    input="Explain quantum entanglement in simple terms.",
    stream=True,
)

for chunk in stream:
    if chunk.event_type == "content.delta":
        if chunk.delta.type == "text":
            print(chunk.delta.text, end="", flush=True)

JavaScript

// Legacy streaming used content.delta
const stream = await client.interactions.create({
    model: 'gemini-3-flash-preview',
    input: 'Explain quantum entanglement in simple terms.',
    stream: true,
});

for await (const chunk of stream) {
    if (chunk.event_type === 'content.delta') {
        if (chunk.delta.type === 'text') {
            process.stdout.write(chunk.delta.text);
        }
    }
}

REST

curl -X POST "https://generativelanguage.googleapis.com/v1beta/interactions?key=$GEMINI_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gemini-3-flash-preview",
    "input": "Explain quantum entanglement in simple terms.",
    "stream": true
  }'

// Response (SSE Lines)
// event: interaction.start
// data: {"id": "int_123", "status": "in_progress"}
//
// event: content.start
// data: {"index": 0, "type": "text"}
//
// event: content.delta
// data: {"delta": {"type": "text", "text": "Quantum entanglement is..."}}
//
// event: content.stop
// data: {"index": 0}
//
// event: interaction.complete
// data: {"id": "int_123", "status": "done", "usage": {"total_tokens": 42}}

之後 (新結構定義)

Python

# Consuming stream and handling new event types
for event in client.interactions.create(
    model="gemini-3-flash-preview",
    input="Tell me a story.",
    stream=True,
):
    if event.type == "step.delta":  # CHANGED: step.delta instead of content.delta
        if event.delta.type == "text":
            print(event.delta.text, end="")

JavaScript

// Consuming stream and handling new event types
const stream = await client.interactions.create({
    model: 'gemini-3-flash-preview',
    input: 'Tell me a story.',
    stream: true,
});

for await (const event of stream) {
    if (event.type === 'step.delta') {  // CHANGED: step.delta instead of content.delta
        if (event.delta.type === 'text') {
            process.stdout.write(event.delta.text);
        }
    }
}

REST

 # Opt-in needed before May 26th
 curl -X POST "https://generativelanguage.googleapis.com/v1beta/interactions?key=$GEMINI_API_KEY" \
   -H "Content-Type: application/json" \
   -H "Accept: text/event-stream" \
   -H "Api-Revision: 2026-05-20" \
   -d '{
     "model": "gemini-3-flash-preview",
     "input": "Tell me a story.",
     "stream": true
   }'

 // Response (SSE Lines)
 // event: interaction.created
 // data: {"interaction": {"id": "int_xyz", "status": "in_progress", "object": "interaction", "model": "gemini-3-flash-preview"}, "event_type": "interaction.created"}
 //
 // event: interaction.in_progress
 // data: {"interaction_id": "int_xyz", "event_type": "interaction.in_progress"}
 //
 // event: step.start
 // data: {"index": 0, "step": {"type": "thought", "signature": "abc123..."}, "event_type": "step.start"}
 //
 // event: step.stop
 // data: {"index": 0, "event_type": "step.stop"}
 //
 // event: step.start
 // data: {"index": 1, "step": {"content": [{"text": "Once upon", "type": "text"}], "type": "model_output"}, "event_type": "step.start"}
 //
 // event: step.delta
 // data: {"index": 1, "delta": {"text": " a time...", "type": "text"}, "event_type": "step.delta"}
 //
 // event: step.stop
 // data: {"type": "step.stop", "index": 1, "status": "done"}
 //
 // event: interaction.completed
 // data: {"type": "interaction.completed", "interaction": {"id": "int_xyz", "status": "completed", "usage": {"prompt_tokens": 10, "completion_tokens": 5, "total_tokens": 15}}} // NEW: Dedicated completion event

無狀態對話記錄

如果您在用戶端手動管理對話記錄 (無狀態用途)，就必須更新先前回合的串連方式。

舊版：開發人員通常會從回覆中收集 outputs 陣列，並在下一個回合中將其傳回 input 欄位。
新結構：您現在應從回應中收集 steps 陣列，並將其傳遞至下一個要求的 input 欄位，將新的使用者輪流步驟附加為 user_input 步驟。

輸出格式設定：`response_format` 變更

更新後的 API 會將所有輸出格式控制項整合為統一的多型 response_format 欄位。這項做法可集中管理頂層的輸出設定，並讓 generation_config 專注於模型行為 (例如溫度參數、Top-P 和思考)。

主要異動

API 會移除 response_mime_type。現在，您可以在 response_format 內為每個格式項目指定 MIME 類型。
response_format 現在是多型物件 (或陣列)。每個項目都有 type 鑑別器 (text、audio、image) 和類型專屬欄位。如要要求多種輸出模態，請傳遞格式項目的陣列。
image_config已從「generation_config」移至「response_format」。 現在，您可以在 response_format 項目中指定圖片輸出設定，例如 aspect_ratio 和 image_size，並使用 "type": "image"。

結構化輸出內容 (JSON)

新結構定義會移除 response_mime_type 欄位。請改為在 response_format 物件中指定 MIME 類型和 JSON 結構定義，並使用 "type": "text"。

之前 (舊版)

Python

interaction = client.interactions.create(
    model="gemini-3-flash-preview",
    input="Summarize this article.",
    response_mime_type="application/json",
    response_format={
        "type": "object",
        "properties": {
            "summary": {"type": "string"}
        }
    },
)

print(interaction.outputs[-1].text)

JavaScript

const interaction = await client.interactions.create({
    model: 'gemini-3-flash-preview',
    input: 'Summarize this article.',
    responseMimeType: 'application/json',
    responseFormat: {
        type: 'object',
        properties: {
            summary: { type: 'string' }
        }
    },
});

console.log(interaction.outputs[-1].text);

REST

curl -X POST "https://generativelanguage.googleapis.com/v1beta/interactions?key=$GEMINI_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gemini-3-flash-preview",
    "input": "Summarize this article.",
    "response_mime_type": "application/json",
    "response_format": {
      "type": "object",
      "properties": {
        "summary": { "type": "string" }
      }
    }
  }'

之後 (新結構定義)

Python

interaction = client.interactions.create(
    model="gemini-3-flash-preview",
    input="Summarize this article.",
    # response_mime_type is removed — specify mime_type inside response_format
    response_format={
        "type": "text",
        "mime_type": "application/json",
        "schema": {
            "type": "object",
            "properties": {
                "summary": {"type": "string"}
            }
        }
    },
)

print(interaction.steps[-1].content[0].text)

JavaScript

const interaction = await client.interactions.create({
    model: 'gemini-3-flash-preview',
    input: 'Summarize this article.',
    // responseMimeType is removed — specify mimeType inside responseFormat
    responseFormat: {
        type: 'text',
        mimeType: 'application/json',
        schema: {
            type: 'object',
            properties: {
                summary: { type: 'string' }
            }
        }
    },
});

console.log(interaction.steps.at(-1).content[0].text);

REST

# Opt-in needed before May 26th
curl -X POST "https://generativelanguage.googleapis.com/v1beta/interactions?key=$GEMINI_API_KEY" \
  -H "Content-Type: application/json" \
  -H "Api-Revision: 2026-05-20" \
  -d '{
    "model": "gemini-3-flash-preview",
    "input": "Summarize this article.",
    "response_format": {
      "type": "text",
      "mime_type": "application/json",
      "schema": {
        "type": "object",
        "properties": {
          "summary": { "type": "string" }
        }
      }
    }
  }'

圖片設定

新結構定義會從 generation_config 中移除 image_config。您現在可以在 "type": "image" 的 response_format 項目中指定圖片輸出設定。

之前 (舊版)

Python

interaction = client.interactions.create(
    model="gemini-3-flash-preview",
    input="Generate an image of a sunset over the ocean.",
    generation_config={
        "image_config": {
            "aspect_ratio": "1:1",
            "image_size": "1K"
        }
    },
)

JavaScript

const interaction = await client.interactions.create({
    model: 'gemini-3-flash-preview',
    input: 'Generate an image of a sunset over the ocean.',
    generationConfig: {
        imageConfig: {
            aspectRatio: '1:1',
            imageSize: '1K'
        }
    },
});

REST

curl -X POST "https://generativelanguage.googleapis.com/v1beta/interactions?key=$GEMINI_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gemini-3-flash-preview",
    "input": "Generate an image of a sunset over the ocean.",
    "generation_config": {
      "image_config": {
        "aspect_ratio": "1:1",
        "image_size": "1K"
      }
    }
  }'

之後 (新結構定義)

Python

interaction = client.interactions.create(
    model="gemini-3-flash-preview",
    input="Generate an image of a sunset over the ocean.",
    # image_config is removed from generation_config — use response_format
    response_format={
        "type": "image",
        "mime_type": "image/jpeg",
        "delivery": "inline",
        "aspect_ratio": "1:1",
        "image_size": "1K"
    },
)

JavaScript

const interaction = await client.interactions.create({
    model: 'gemini-3-flash-preview',
    input: 'Generate an image of a sunset over the ocean.',
    // imageConfig is removed from generationConfig — use responseFormat
    responseFormat: {
        type: 'image',
        mimeType: 'image/jpeg',
        delivery: 'inline',
        aspectRatio: '1:1',
        imageSize: '1K'
    },
});

REST

# Opt-in needed before May 26th
curl -X POST "https://generativelanguage.googleapis.com/v1beta/interactions?key=$GEMINI_API_KEY" \
  -H "Content-Type: application/json" \
  -H "Api-Revision: 2026-05-20" \
  -d '{
    "model": "gemini-3-flash-preview",
    "input": "Generate an image of a sunset over the ocean.",
    "response_format": {
      "type": "image",
      "mime_type": "image/jpeg",
      "delivery": "inline",
      "aspect_ratio": "1:1",
      "image_size": "1K"
    }
  }'

如要要求多種輸出模態 (例如同時輸出文字和音訊)，請將格式項目陣列傳遞至 response_format，而非單一物件。

如何遷移至新結構定義

SDK 使用者

升級至最新版 SDK (Python ≥2.0.0、JavaScript ≥2.0.0)。SDK 會自動選擇採用新架構，您只需要更新讀取回應的方式 (請參閱上方的範例)，不必變更任何程式碼。請注意，這些 SDK 版本僅支援新結構定義。在 2026 年 6 月 8 日移除舊版結構定義前，舊版 SDK (Python 1.x.x、JavaScript 1.x.x) 仍可繼續運作。

REST API 使用者

在要求中加入 Api-Revision: 2026-05-20 標頭，即可立即選擇採用新結構定義。5 月 26 日後，所有要求都會預設使用新結構定義。您可以使用 Api-Revision: 2026-05-07 暫時停用，但 6 月 8 日後，API 就會永久移除舊版結構定義。

時間軸

日期	階段	SDK 使用者	REST API 使用者
5 月 7 日	啟用	推出新版 SDK (Python ≥2.0.0、JS ≥2.0.0)。升級即可自動取得新結構定義。	新增 `Api-Revision: 2026-05-20` 標頭即可選擇加入。預設值仍為舊版。
5 月 26 日	預設翻轉	如果已升級，則無須採取任何行動。舊版 SDK (Python 1.x.x、JS 1.x.x) 仍可運作，但會傳回舊版的回應。	新結構定義現在為預設結構定義。傳送 `Api-Revision: 2026-05-07` 標頭即可停用。
6 月 8 日	日落	Python 1.x.x 和 JS 1.x.x SDK 版本會中斷 Interactions API 呼叫。	已移除 Interactions API 的舊版結構定義。系統會忽略 `Api-Revision` 標頭。

遷移檢查清單

步驟結構定義 (`steps`)

更新程式碼，從 steps 陣列而非 outputs 讀取回應內容。查看範例。
確認程式碼可處理 user_input 和 model_output 步驟類型。查看範例。
(函式呼叫) 更新程式碼，在 steps 陣列中找出 function_call 步驟。查看範例。
(伺服器端工具) 更新程式碼，處理工具專屬步驟 (例如 google_search_call、google_search_result)。查看範例。
(無狀態記錄) 更新記錄管理，在下一個要求的 input 欄位中傳遞 steps 陣列。查看詳細資料。
(僅限串流) 更新用戶端，監聽新的 SSE 事件類型 (interaction.created、step.delta 等)。查看範例。

輸出格式設定 (`response_format`)

將 response_format 內的 mime_type 欄位替換為 response_mime_type。查看範例。
將現有的 response_format JSON 結構定義包裝在 {"type": "text", "schema": ...} 物件中。查看範例。
(圖像生成) 將 image_config 從 generation_config 移至 response_format 中的 {"type": "image", ...} 項目。查看範例。
(多模態) 要求多個輸出模態時，請將 response_format 從單一物件轉換為陣列。

Interactions API：重大變更遷移指南 (2026 年 5 月)

核心異動：outputs 改為 steps

基本輸入/輸出 (一元)

之前 (舊版)

Python

JavaScript

REST

之後 (新結構定義)

Python

JavaScript

REST

函式呼叫

之前 (舊版)

Python

JavaScript

REST

之後 (新結構定義)

Python

JavaScript

REST

伺服器端工具

之前 (舊版)

Python

JavaScript

REST

之後 (新結構定義)

Python

JavaScript

REST

串流

新事件類型

已淘汰的事件類型

範例

之前 (舊版)

Python

JavaScript

REST

之後 (新結構定義)

Python

JavaScript

REST

無狀態對話記錄

輸出格式設定：response_format 變更

主要異動

結構化輸出內容 (JSON)

之前 (舊版)

Python

JavaScript

REST

之後 (新結構定義)

Python

JavaScript

REST

圖片設定

之前 (舊版)

Python

JavaScript

REST

之後 (新結構定義)

Python

JavaScript

REST

如何遷移至新結構定義

SDK 使用者

REST API 使用者

時間軸

遷移檢查清單

步驟結構定義 (steps)

輸出格式設定 (response_format)

核心異動：`outputs` 改為 `steps`

輸出格式設定：`response_format` 變更

步驟結構定義 (`steps`)

輸出格式設定 (`response_format`)