Dateisuche | OpenAI API

Die Dateisuche ist ein Tool der Responses API. Damit können Modelle Informationen aus einer Wissensbasis zuvor hochgeladener Dateien mithilfe semantischer Suche und Stichwortsuche abrufen. Indem du Vektorspeicher erstellst und Dateien darin hochlädst, kannst du das vorhandene Wissen der Modelle ergänzen: Du gibst ihnen Zugriff auf diese Wissensbasen, auch vector_stores genannt.

Mehr darüber, wie Vektorspeicher und die semantische Suche funktionieren, erfährst du in unserem Leitfaden zum Informationsabruf.

Dieses Tool wird von OpenAI gehostet und verwaltet. Du musst daher keinen eigenen Code für seine Ausführung implementieren. Wenn das Modell entscheidet, es zu verwenden, ruft es das Tool automatisch auf, ruft Informationen aus deinen Dateien ab und gibt ein Ergebnis zurück.

Verwendung

Bevor du die Dateisuche mit der Responses API verwenden kannst, musst du eine Wissensbasis in einem Vektorspeicher einrichten und Dateien darin hochladen.

Gehe wie folgt vor, um einen Vektorspeicher zu erstellen und eine Datei darin hochzuladen. Du kannst diese Beispieldatei verwenden oder eine eigene hochladen.

Die Datei in die File API hochladen

Eine Datei hochladen

Python

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34import fs from "fs";
import OpenAI from "openai";
const openai = new OpenAI();

async function createFile(filePath) {
  let result;
  if (filePath.startsWith("http://") || filePath.startsWith("https://")) {
    // Download the file content from the URL
    const res = await fetch(filePath);
    const buffer = await res.arrayBuffer();
    const urlParts = filePath.split("/");
    const fileName = urlParts[urlParts.length - 1];
    const file = new File([buffer], fileName);
    result = await openai.files.create({
      file: file,
      purpose: "assistants",
    });
  } else {
    // Handle local file path
    const fileContent = fs.createReadStream(filePath);
    result = await openai.files.create({
      file: fileContent,
      purpose: "assistants",
    });
  }
  return result.id;
}

// Replace with your own file path or URL
const fileId = await createFile(
  "https://cdn.openai.com/API/docs/deep_research_blog.pdf"
);

console.log(fileId);

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32from io import BytesIO

import requests
from openai import OpenAI

client = OpenAI()


def create_file(client, file_path):
    if file_path.startswith(("http://", "https://")):
        response = requests.get(file_path, timeout=30)
        response.raise_for_status()
        file_content = BytesIO(response.content)
        file_name = file_path.rsplit("/", 1)[-1]
        result = client.files.create(
            file=(file_name, file_content),
            purpose="assistants",
        )
    else:
        with open(file_path, "rb") as file_content:
            result = client.files.create(
                file=file_content,
                purpose="assistants",
            )
    return result.id


file_id = create_file(
    client,
    "https://cdn.openai.com/API/docs/deep_research_blog.pdf",
)
print(file_id)

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27package main

import (
	"context"
	"fmt"
	"os"

	"github.com/openai/openai-go/v3"
)

func main() {
	file, err := os.Open("customer_policies.txt")
	if err != nil {
		panic(err)
	}
	defer file.Close()

	client := openai.NewClient()
	result, err := client.Files.New(context.Background(), openai.FileNewParams{
		File:    openai.File(file, "customer_policies.txt", "text/plain"),
		Purpose: openai.FilePurposeAssistants,
	})
	if err != nil {
		panic(err)
	}
	fmt.Println(result.ID)
}

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16import com.openai.client.OpenAIClient;
import com.openai.client.okhttp.OpenAIOkHttpClient;
import com.openai.models.files.FileCreateParams;
import com.openai.models.files.FilePurpose;
import java.nio.file.Path;

var file =
    client
        .files()
        .create(
            FileCreateParams.builder()
                .file(Path.of(System.getenv("OPENAI_EXAMPLE_FILE_PATH")))
                .purpose(FilePurpose.USER_DATA)
                .build());

System.out.println(file.id());

1
2
3
4
5
6
7require "openai"
require "pathname"

client = OpenAI::Client.new
file = Pathname("customer_policies.txt")
uploaded = client.files.create(file: file, purpose: :user_data)
puts(uploaded.id)

Einen Vektorspeicher erstellen

Python

const vectorStore = await openai.vectorStores.create({
  name: "knowledge_base",
});
console.log(vectorStore.id);

vector_store = client.vector_stores.create(name="knowledge_base")
print(vector_store.id)

package main

import (
	"context"
	"fmt"

	"github.com/openai/openai-go/v3"
)

func main() {
	client := openai.NewClient()
	vectorStore, err := client.VectorStores.New(context.Background(), openai.VectorStoreNewParams{
		Name: openai.String("knowledge_base"),
	})
	if err != nil {
		panic(err)
	}
	fmt.Println(vectorStore.ID)
}

import com.openai.client.OpenAIClient;
import com.openai.client.okhttp.OpenAIOkHttpClient;
import com.openai.models.vectorstores.VectorStoreCreateParams;

var store =
    client
        .vectorStores()
        .create(VectorStoreCreateParams.builder().name("Product docs").build());

System.out.println(store.id());

require "openai"

client = OpenAI::Client.new
store = client.vector_stores.create(name: "Product docs")
puts(store.id)

Die Datei zum Vektorspeicher hinzufügen

Eine Datei zu einem Vektorspeicher hinzufügen

Python

1
2
3
4// Use vectorStore and fileId from the earlier create and upload steps.
await openai.vectorStores.files.create(vectorStore.id, {
  file_id: fileId,
});

1
2
3
4
5result = client.vector_stores.files.create(
    vector_store_id=vector_store.id,
    file_id=file_id,
)
print(result)

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19package main

import (
	"context"
	"fmt"

	"github.com/openai/openai-go/v3"
)

func main() {
	client := openai.NewClient()
	file, err := client.VectorStores.Files.New(context.Background(), "<vector_store_id>", openai.VectorStoreFileNewParams{
		FileID: "file_abc123",
	})
	if err != nil {
		panic(err)
	}
	fmt.Println(file.ID)
}

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15import com.openai.client.OpenAIClient;
import com.openai.client.okhttp.OpenAIOkHttpClient;
import com.openai.models.vectorstores.files.FileCreateParams;

String vectorStoreId = "<vector_store_id>";

String fileId = "file_abc123";

var file =
    client
        .vectorStores()
        .files()
        .create(vectorStoreId, FileCreateParams.builder().fileId(fileId).build());

System.out.println(file.id());

1
2
3
4
5require "openai"

client = OpenAI::Client.new
file = client.vector_stores.files.create("<vector_store_id>", file_id: "file_abc123")
puts(file.id)

Status prüfen

Führe diesen Code so lange aus, bis die Datei verwendet werden kann, also bis der Status completed lautet.

Status prüfen

Python

// Use vectorStore from the earlier create step.
const result = await openai.vectorStores.files.list(vectorStore.id);
console.log(result);

result = client.vector_stores.files.list(vector_store_id=vector_store.id)
print(result)

package main

import (
	"context"
	"fmt"

	"github.com/openai/openai-go/v3"
)

func main() {
	client := openai.NewClient()
	files, err := client.VectorStores.Files.List(context.Background(), "<vector_store_id>", openai.VectorStoreFileListParams{})
	if err != nil {
		panic(err)
	}
	fmt.Println(files.Data)
}

import com.openai.client.OpenAIClient;
import com.openai.client.okhttp.OpenAIOkHttpClient;

String vectorStoreId = "<vector_store_id>";

System.out.println(client.vectorStores().files().list(vectorStoreId).data());

require "openai"

client = OpenAI::Client.new
files = client.vector_stores.files.list("<vector_store_id>")
puts(files.data&.map(&:status))

Sobald deine Wissensbasis eingerichtet ist, kannst du das Tool file_search in die Liste der für das Modell verfügbaren Tools aufnehmen und die Liste der zu durchsuchenden Vektorspeicher angeben.

Tool zur Dateisuche

Python

1
2
3
4
5
6
7
8
9
10
11
12
13
14import OpenAI from "openai";
const openai = new OpenAI();

const response = await openai.responses.create({
  model: "gpt-6-astra",
  input: "What is deep research by OpenAI?",
  tools: [
    {
      type: "file_search",
      vector_store_ids: ["<vector_store_id>"],
    },
  ],
});
console.log(response);

1
2
3
4
5
6
7
8
9
10from openai import OpenAI

client = OpenAI()

response = client.responses.create(
    model="gpt-6-astra",
    input="What is deep research by OpenAI?",
    tools=[{"type": "file_search", "vector_store_ids": ["<vector_store_id>"]}],
)
print(response)

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22package main

import (
	"context"
	"fmt"

	"github.com/openai/openai-go/v3"
	"github.com/openai/openai-go/v3/responses"
)

func main() {
	client := openai.NewClient()
	response, err := client.Responses.New(context.Background(), responses.ResponseNewParams{
		Model: "gpt-6-astra",
		Input: responses.ResponseNewParamsInputUnion{OfString: openai.String("What is deep research by OpenAI?")},
		Tools: []responses.ToolUnionParam{responses.ToolParamOfFileSearch([]string{"<vector_store_id>"})},
	})
	if err != nil {
		panic(err)
	}
	fmt.Println(response)
}

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19import com.openai.client.OpenAIClient;
import com.openai.client.okhttp.OpenAIOkHttpClient;
import com.openai.models.responses.ResponseCreateParams;
import java.util.List;

String vectorStoreId = "<vector_store_id>";

ResponseCreateParams params =
    ResponseCreateParams.builder()
        .model("gpt-6-astra")
        .input("What is deep research by OpenAI?")
        .addFileSearchTool(List.of(vectorStoreId))
        .build();

client.responses().create(params).output().stream()
    .flatMap(item -> item.message().stream())
    .flatMap(message -> message.content().stream())
    .flatMap(content -> content.outputText().stream())
    .forEach(text -> System.out.println(text.text()));

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18using OpenAI.Responses;
#pragma warning disable OPENAI001

string key = Environment.GetEnvironmentVariable("OPENAI_API_KEY")!;
string vectorStoreId = "<vector_store_id>";
ResponsesClient client = new(key);

CreateResponseOptions options = new() { Model = "gpt-6-astra" };
options.Tools.Add(
    ResponseTool.CreateFileSearchTool([vectorStoreId])
);
options.InputItems.Add(
    ResponseItem.CreateUserMessageItem("What is deep research by OpenAI?")
);

ResponseResult response = await client.CreateResponseAsync(options);

Console.WriteLine(response.GetOutputText());

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16require "openai"

openai = OpenAI::Client.new

response = openai.responses.create(
  model: "gpt-6-astra",
  input: "What is deep research by OpenAI?",
  tools: [
    {
      type: "file_search",
      vector_store_ids: ["<vector_store_id>"]
    }
  ]
)

puts(response)

Wenn das Modell dieses Tool aufruft, erhältst du eine Antwort mit mehreren Ausgabeelementen:

Ein Ausgabeelement vom Typ file_search_call, das die ID des Dateisuche-Aufrufs enthält.
Ein Ausgabeelement vom Typ message, das die Antwort des Modells samt Quellenverweisen auf die Dateien enthält.

Antwort der Dateisuche

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48{
  "output": [
    {
      "type": "file_search_call",
      "id": "fs_67c09ccea8c48191ade9367e3ba71515",
      "status": "completed",
      "queries": ["What is deep research?"],
      "search_results": null
    },
    {
      "id": "msg_67c09cd3091c819185af2be5d13d87de",
      "type": "message",
      "role": "assistant",
      "content": [
        {
          "type": "output_text",
          "text": "Deep research is a sophisticated capability that allows for extensive inquiry and synthesis of information across various domains. It is designed to conduct multi-step research tasks, gather data from multiple online sources, and provide comprehensive reports similar to what a research analyst would produce. This functionality is particularly useful in fields requiring detailed and accurate information...",
          "annotations": [
            {
              "type": "file_citation",
              "index": 992,
              "file_id": "file-2dtbBZdjtDKS8eqWxqbgDi",
              "filename": "deep_research_blog.pdf"
            },
            {
              "type": "file_citation",
              "index": 992,
              "file_id": "file-2dtbBZdjtDKS8eqWxqbgDi",
              "filename": "deep_research_blog.pdf"
            },
            {
              "type": "file_citation",
              "index": 1176,
              "file_id": "file-2dtbBZdjtDKS8eqWxqbgDi",
              "filename": "deep_research_blog.pdf"
            },
            {
              "type": "file_citation",
              "index": 1176,
              "file_id": "file-2dtbBZdjtDKS8eqWxqbgDi",
              "filename": "deep_research_blog.pdf"
            }
          ]
        }
      ]
    }
  ]
}

Informationsabruf anpassen

Anzahl der Ergebnisse begrenzen

Wenn du das Tool zur Dateisuche mit der Responses API verwendest, kannst du festlegen, wie viele Ergebnisse aus den Vektorspeichern abgerufen werden sollen. Dadurch lassen sich Tokenverbrauch und Latenz reduzieren, allerdings kann die Qualität der Antworten darunter leiden.

Anzahl der Ergebnisse begrenzen

Python

1
2
3
4
5
6
7
8
9
10
11
12const response = await openai.responses.create({
  model: "gpt-6-astra",
  input: "What is deep research by OpenAI?",
  tools: [
    {
      type: "file_search",
      vector_store_ids: ["<vector_store_id>"],
      max_num_results: 2,
    },
  ],
});
console.log(response);

1
2
3
4
5
6
7
8
9
10
11
12response = client.responses.create(
    model="gpt-6-astra",
    input="What is deep research by OpenAI?",
    tools=[
        {
            "type": "file_search",
            "vector_store_ids": ["<vector_store_id>"],
            "max_num_results": 2,
        }
    ],
)
print(response)

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24package main

import (
	"context"
	"fmt"

	"github.com/openai/openai-go/v3"
	"github.com/openai/openai-go/v3/responses"
)

func main() {
	client := openai.NewClient()
	tool := responses.ToolParamOfFileSearch([]string{"<vector_store_id>"})
	tool.OfFileSearch.MaxNumResults = openai.Int(2)
	response, err := client.Responses.New(context.Background(), responses.ResponseNewParams{
		Model: "gpt-6-astra",
		Input: responses.ResponseNewParamsInputUnion{OfString: openai.String("What is deep research by OpenAI?")},
		Tools: []responses.ToolUnionParam{tool},
	})
	if err != nil {
		panic(err)
	}
	fmt.Println(response)
}

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20import com.openai.client.OpenAIClient;
import com.openai.client.okhttp.OpenAIOkHttpClient;
import com.openai.models.responses.FileSearchTool;
import com.openai.models.responses.ResponseCreateParams;

String vectorStoreId = "<vector_store_id>";

ResponseCreateParams params =
    ResponseCreateParams.builder()
        .model("gpt-6-astra")
        .input("What is deep research by OpenAI?")
        .addTool(
            FileSearchTool.builder().addVectorStoreId(vectorStoreId).maxNumResults(2).build())
        .build();

client.responses().create(params).output().stream()
    .flatMap(item -> item.message().stream())
    .flatMap(message -> message.content().stream())
    .flatMap(content -> content.outputText().stream())
    .forEach(text -> System.out.println(text.text()));

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18using OpenAI.Responses;
#pragma warning disable OPENAI001

string key = Environment.GetEnvironmentVariable("OPENAI_API_KEY")!;
// Replace this illustrative ID with your vector store ID.
string vectorStoreId = "<vector_store_id>";
ResponsesClient client = new(key);

CreateResponseOptions options = new() { Model = "gpt-6-astra" };
options.Tools.Add(
    ResponseTool.CreateFileSearchTool([vectorStoreId], maxResultCount: 2)
);
options.InputItems.Add(
    ResponseItem.CreateUserMessageItem("What is deep research by OpenAI?")
);

ResponseResult response = await client.CreateResponseAsync(options);
Console.WriteLine(response.GetOutputText());

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17require "openai"

client = OpenAI::Client.new

response = client.responses.create(
  model: "gpt-6-astra",
  input: "What is deep research by OpenAI?",
  tools: [
    {
      type: :file_search,
      vector_store_ids: ["<vector_store_id>"],
      max_num_results: 2
    }
  ]
)

puts(response)

Suchergebnisse in die Antwort aufnehmen

Im Ausgabetext siehst du zwar Anmerkungen mit Verweisen auf Dateien, der Dateisuche-Aufruf gibt jedoch standardmäßig keine Suchergebnisse zurück.

Um Suchergebnisse in die Antwort aufzunehmen, kannst du beim Erstellen der Antwort den Parameter include verwenden.

Suchergebnisse aufnehmen

Python

1
2
3
4
5
6
7
8
9
10
11
12const response = await openai.responses.create({
  model: "gpt-6-astra",
  input: "What is deep research by OpenAI?",
  tools: [
    {
      type: "file_search",
      vector_store_ids: ["<vector_store_id>"],
    },
  ],
  include: ["file_search_call.results"],
});
console.log(response);

1
2
3
4
5
6
7
8
9
10
11
12response = client.responses.create(
    model="gpt-6-astra",
    input="What is deep research by OpenAI?",
    tools=[
        {
            "type": "file_search",
            "vector_store_ids": ["<vector_store_id>"],
        }
    ],
    include=["file_search_call.results"],
)
print(response)

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23package main

import (
	"context"
	"fmt"

	"github.com/openai/openai-go/v3"
	"github.com/openai/openai-go/v3/responses"
)

func main() {
	client := openai.NewClient()
	response, err := client.Responses.New(context.Background(), responses.ResponseNewParams{
		Model:   "gpt-6-astra",
		Input:   responses.ResponseNewParamsInputUnion{OfString: openai.String("What is deep research by OpenAI?")},
		Tools:   []responses.ToolUnionParam{responses.ToolParamOfFileSearch([]string{"<vector_store_id>"})},
		Include: []responses.ResponseIncludable{responses.ResponseIncludableFileSearchCallResults},
	})
	if err != nil {
		panic(err)
	}
	fmt.Println(response)
}

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21import com.openai.client.OpenAIClient;
import com.openai.client.okhttp.OpenAIOkHttpClient;
import com.openai.models.responses.ResponseCreateParams;
import com.openai.models.responses.ResponseIncludable;
import java.util.List;

String vectorStoreId = "<vector_store_id>";

ResponseCreateParams params =
    ResponseCreateParams.builder()
        .model("gpt-6-astra")
        .input("What is deep research by OpenAI?")
        .addInclude(ResponseIncludable.of("file_search_call.results"))
        .addFileSearchTool(List.of(vectorStoreId))
        .build();

client.responses().create(params).output().stream()
    .flatMap(item -> item.fileSearchCall().stream())
    .flatMap(call -> call.results().stream())
    .flatMap(List::stream)
    .forEach(System.out::println);

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23using OpenAI.Responses;
#pragma warning disable OPENAI001

string key = Environment.GetEnvironmentVariable("OPENAI_API_KEY")!;
// Replace this illustrative ID with your vector store ID.
string vectorStoreId = "<vector_store_id>";
ResponsesClient client = new(key);

CreateResponseOptions options = new() { Model = "gpt-6-astra" };
options.Tools.Add(ResponseTool.CreateFileSearchTool([vectorStoreId]));
options.IncludedProperties.Add(IncludedResponseProperty.FileSearchCallResults);
options.InputItems.Add(
    ResponseItem.CreateUserMessageItem("What is deep research by OpenAI?")
);

ResponseResult response = await client.CreateResponseAsync(options);
foreach (FileSearchCallResponseItem search in response.OutputItems.OfType<FileSearchCallResponseItem>())
{
    foreach (FileSearchCallResult result in search.Results)
    {
        Console.WriteLine($"{result.Filename}: {result.Text}");
    }
}

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17require "openai"

client = OpenAI::Client.new

response = client.responses.create(
  model: "gpt-6-astra",
  input: "What is deep research by OpenAI?",
  include: ["file_search_call.results"],
  tools: [
    {
      type: :file_search,
      vector_store_ids: ["<vector_store_id>"]
    }
  ]
)

puts(response)

Nach Metadaten filtern

Du kannst die Suchergebnisse anhand der Metadaten der Dateien filtern. Weitere Informationen findest du in unserem Leitfaden zum Informationsabruf. Dort erfährst du:

Wie du Attribute für Dateien in Vektorspeichern festlegst
Wie du Filter definierst

Nach Metadaten filtern

Python

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16const response = await openai.responses.create({
  model: "gpt-6-astra",
  input: "What is deep research by OpenAI?",
  tools: [
    {
      type: "file_search",
      vector_store_ids: ["<vector_store_id>"],
      filters: {
        type: "in",
        key: "category",
        value: ["blog", "announcement"],
      },
    },
  ],
});
console.log(response);

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16response = client.responses.create(
    model="gpt-6-astra",
    input="What is deep research by OpenAI?",
    tools=[
        {
            "type": "file_search",
            "vector_store_ids": ["<vector_store_id>"],
            "filters": {
                "type": "in",
                "key": "category",
                "value": ["blog", "announcement"],
            },
        }
    ],
)
print(response)

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34package main

import (
	"context"
	"fmt"

	"github.com/openai/openai-go/v3"
	"github.com/openai/openai-go/v3/responses"
	"github.com/openai/openai-go/v3/shared"
)

func main() {
	client := openai.NewClient()
	tool := responses.ToolParamOfFileSearch([]string{"<vector_store_id>"})
	tool.OfFileSearch.Filters = responses.FileSearchToolFiltersUnionParam{
		OfComparisonFilter: &shared.ComparisonFilterParam{
			Type: shared.ComparisonFilterTypeIn,
			Key:  "category",
			Value: shared.ComparisonFilterValueUnionParam{OfComparisonFilterValueArray: []shared.ComparisonFilterValueArrayItemUnionParam{
				{OfString: openai.String("blog")},
				{OfString: openai.String("announcement")},
			}},
		},
	}
	response, err := client.Responses.New(context.Background(), responses.ResponseNewParams{
		Model: "gpt-6-astra",
		Input: responses.ResponseNewParamsInputUnion{OfString: openai.String("What is deep research by OpenAI?")},
		Tools: []responses.ToolUnionParam{tool},
	})
	if err != nil {
		panic(err)
	}
	fmt.Println(response)
}

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35import com.openai.client.OpenAIClient;
import com.openai.client.okhttp.OpenAIOkHttpClient;
import com.openai.models.ComparisonFilter;
import com.openai.models.responses.FileSearchTool;
import com.openai.models.responses.ResponseCreateParams;
import java.util.List;

String vectorStoreId = "<vector_store_id>";

ResponseCreateParams params =
    ResponseCreateParams.builder()
        .model("gpt-6-astra")
        .input("What is deep research by OpenAI?")
        .addTool(
            FileSearchTool.builder()
                .addVectorStoreId(vectorStoreId)
                .filters(
                    ComparisonFilter.builder()
                        .type(ComparisonFilter.Type.IN)
                        .key("category")
                        .valueOfComparisonFilterValueItems(
                            List.of(
                                ComparisonFilter.Value.ComparisonFilterValueItem.ofString(
                                    "blog"),
                                ComparisonFilter.Value.ComparisonFilterValueItem.ofString(
                                    "announcement")))
                        .build())
                .build())
        .build();

client.responses().create(params).output().stream()
    .flatMap(item -> item.message().stream())
    .flatMap(message -> message.content().stream())
    .flatMap(content -> content.outputText().stream())
    .forEach(text -> System.out.println(text.text()));

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23using OpenAI.Responses;
#pragma warning disable OPENAI001

string key = Environment.GetEnvironmentVariable("OPENAI_API_KEY")!;
// Replace this illustrative ID with your vector store ID.
string vectorStoreId = "<vector_store_id>";
ResponsesClient client = new(key);

BinaryData filters = BinaryData.FromString(
    """
    { "type": "in", "key": "category", "value": ["blog", "announcement"] }
    """
);
CreateResponseOptions options = new() { Model = "gpt-6-astra" };
options.Tools.Add(
    ResponseTool.CreateFileSearchTool([vectorStoreId], filters: filters)
);
options.InputItems.Add(
    ResponseItem.CreateUserMessageItem("What is deep research by OpenAI?")
);

ResponseResult response = await client.CreateResponseAsync(options);
Console.WriteLine(response.GetOutputText());

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21require "openai"

client = OpenAI::Client.new

response = client.responses.create(
  model: "gpt-6-astra",
  input: "What is deep research by OpenAI?",
  tools: [
    {
      type: :file_search,
      vector_store_ids: ["<vector_store_id>"],
      filters: {
        type: :in,
        key: "category",
        value: ["blog", "announcement"]
      }
    }
  ]
)

puts(response)

Unterstützte Dateien

Für MIME-Typen mit text/ muss die Zeichenkodierung utf-8, utf-16 oder ascii sein.

Dateiformat	MIME-Typ
`.c`	`text/x-c`
`.cpp`	`text/x-c++`
`.cs`	`text/x-csharp`
`.css`	`text/css`
`.doc`	`application/msword`
`.docx`	`application/vnd.openxmlformats-officedocument.wordprocessingml.document`
`.go`	`text/x-golang`
`.html`	`text/html`
`.java`	`text/x-java`
`.js`	`text/javascript`
`.json`	`application/json`
`.md`	`text/markdown`
`.pdf`	`application/pdf`
`.php`	`text/x-php`
`.pptx`	`application/vnd.openxmlformats-officedocument.presentationml.presentation`
`.py`	`text/x-python`
`.py`	`text/x-script.python`
`.rb`	`text/x-ruby`
`.sh`	`application/x-sh`
`.tex`	`text/x-tex`
`.ts`	`application/typescript`
`.txt`	`text/plain`

Hinweise zur Nutzung

API-Verfügbarkeit	Ratenlimits	Hinweise
Responses Chat Completions Assistants	Stufe 1 100 RPM Stufe 2 und 3 500 RPM Stufe 4 und 5 1.000 RPM	Preise ZDR und Datenresidenz

API-Verfügbarkeit

Ratenlimits

Hinweise

Responses

Chat Completions

Assistants

Stufe 1
100 RPM

Stufe 2 und 3
500 RPM

Stufe 4 und 5
1.000 RPM

Preise
ZDR und Datenresidenz

Vorschläge

Erste Schritte

Grundlegende Konzepte

SDKs und CLI

Ressourcen

Ältere APIs

Modell auswählen

Text und Code

Prompts formulieren

Reasoning-Aufwand

Bilder und Video

Echtzeit und Audio

Spezialisierte Modelle

Agents API

Agents SDK

ChatKit

Suche und Informationsabruf

Tools und Daten anbinden

Arbeitsabläufe mit Tools erstellen

Computer und Code

Medien

GPT-Live

Realtime API

Anwendungen mit Sprachfunktionen entwickeln

Verbindungen

Audioverarbeitung

In Produktion gehen

Leistung und Qualität

Kosten und Durchsatz

Sicherheit und Governance

Infrastruktur und Zugriff

Betrieb

Grundlegende Konzepte

Plan

Build

Test and publish

Conversion specs

Leitfäden

Ressourcen

Erste Schritte

Leitfäden

File Upload

API

Measurement

Advertiser API

API-Referenz

Erste Schritte

Grundlagen

Entdecken

Verfügbar auf

Versionen

Arbeitsabläufe

Funktionen

Referenz

Anpassung

Konfigurationsdatei

Agentenkonfiguration

ChatGPT und Codex erweitern

Linux

Windows

Arbeitsabläufe für die Entwicklung

Erweitern und automatisieren

Umgebungen

Mit Codex entwickeln

Integrationen von Drittanbietern

Referenz

Berechtigungen

Codex Security

Cybersicherheit

Erste Schritte

ChatGPT Work

Identität und Authentifizierung

Workspace-Zugriff, Richtlinien und Modelle

Kontrollen für Plug-ins und Konnektoren

Nutzung, Governance und Compliance

Bereitstellung und Modellanbieter

Community

Blog

Community

Blog