超快速模式是 OpenAI API 中速度最快的服务层级,支持 GPT-6.2 Astra、GPT-6.2 Sol、GPT-6 Astra 和 GPT-6.1 Sol。如果您的任务值得为更快的响应速度支付更高的费用,可以使用此模式。
要在 ChatGPT 工作或 Codex 中使用超快速模式,请在编辑器中选择速度。
我们强烈建议使用 WebSocket,尤其是在智能体应用需要短时间内连续多次调用工具时。如果不使用持久连接,网络开销可能会抵消一部分延迟优势。
所有 API 用户都可以在 GPT-6 Astra 和 GPT-6.1 Sol 上使用超快速模式。
超快速模式的速率限制独立于标准模式和快速模式。增加流量前,请先查看您所在组织的速率限制。
如果您的组织有对接的 OpenAI 客户团队,请联系该团队申请提高速率限制。
将 model 设为 gpt-6.2-astra、gpt-6.2-sol、gpt-6-astra 或 gpt-6.1-sol。在每个 response.create 事件中,将 service_tier 设为 ultrafast。
请按照建立连接并创建响应中的说明,安装您所用编程语言的 SDK 和 WebSocket 依赖项。运行示例前,请在环境中设置 OPENAI_API_KEY。
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
49
50
51
52
53// Install: npm install openai ws
// Set OPENAI_API_KEY in your environment.
import OpenAI from "openai";
import { ResponsesWS } from "openai/resources/responses/ws";
const client = new OpenAI();
const ws = new ResponsesWS(client);
let previousResponseId = null;
try {
for (const input of [
"Explain why the sky is blue in one sentence.",
"Now explain why sunsets look red.",
]) {
let completed = false;
const events = ws.stream();
ws.send({
type: "response.create",
model: "gpt-6-astra",
service_tier: "ultrafast",
previous_response_id: previousResponseId,
input,
});
for await (const event of events) {
if (event.type === "error") throw event.error;
if (event.type !== "message") continue;
const message = event.message;
if (message.type === "response.output_text.delta") {
process.stdout.write(message.delta);
} else if (message.type === "response.completed") {
previousResponseId = message.response.id;
completed = true;
console.log();
break;
} else if (
message.type === "response.failed" ||
message.type === "response.incomplete"
) {
throw new Error(JSON.stringify(message));
}
}
if (!completed) {
throw new Error("Connection closed before the response finished.");
}
}
} finally {
ws.close();
}
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# Install: pip install --upgrade "openai[realtime]"
# Set OPENAI_API_KEY in your environment.
from openai import OpenAI
client = OpenAI()
previous_response_id: str | None = None
prompts = [
"Explain why the sky is blue in one sentence.",
"Now explain why sunsets look red.",
]
with client.responses.connect() as connection:
for prompt in prompts:
connection.response.create(
model="gpt-6-astra",
service_tier="ultrafast",
store=False, # Keep this conversation on the current connection.
previous_response_id=previous_response_id,
input=prompt,
)
for event in connection:
if event.type == "response.output_text.delta":
print(event.delta, end="", flush=True)
elif event.type == "response.completed":
previous_response_id = event.response.id
print()
break
elif event.type in {"response.failed", "response.incomplete", "error"}:
raise RuntimeError(event.to_json())
else:
raise RuntimeError("Connection closed before the response finished.")
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
49
50
51
52
53
54
55
56package main
import (
"context"
"fmt"
"time"
"github.com/openai/openai-go/v3"
"github.com/openai/openai-go/v3/responses"
)
func main() {
ctx, cancel := context.WithTimeout(context.Background(), 2*time.Minute)
defer cancel()
client := openai.NewClient()
conn, err := client.Responses.Connect(ctx, responses.ResponseConnectionOptions{})
if err != nil {
panic(err)
}
defer conn.Close()
previousResponseID := ""
for _, prompt := range []string{
"Explain why the sky is blue in one sentence.",
"Now explain why sunsets look red.",
} {
request := responses.ResponsesClientEventResponseCreateParam{
Model: "gpt-6-astra",
ServiceTier: "ultrafast",
Input: responses.ResponsesClientEventResponseCreateInputUnionParam{
OfString: openai.String(prompt),
},
}
if previousResponseID != "" {
request.PreviousResponseID = openai.String(previousResponseID)
}
if err := conn.Create(ctx, request); err != nil {
panic(err)
}
// FinalResponse reads events until the response finishes.
response, err := conn.FinalResponse(ctx)
if err != nil {
panic(err)
}
if response.Status != responses.ResponseStatusCompleted {
panic(fmt.Sprintf("Response ended with status %s", response.Status))
}
previousResponseID = response.ID
fmt.Println(response.OutputText())
}
}
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
41import com.openai.client.okhttp.OpenAIOkHttpClient;
import com.openai.models.responses.ResponseStatus;
import com.openai.models.responses.ResponsesClientEvent;
import java.util.List;
var client = OpenAIOkHttpClient.fromEnv();
try (var connection = client.responses().connect()) {
String previousResponseId = null;
for (var prompt :
List.of(
"Explain why the sky is blue in one sentence.",
"Now explain why sunsets look red.")) {
var request =
ResponsesClientEvent.ResponseCreate.builder()
.model("gpt-6-astra")
.serviceTier(ResponsesClientEvent.ResponseCreate.ServiceTier.ULTRAFAST)
.input(prompt);
if (previousResponseId != null) {
request.previousResponseId(previousResponseId);
}
connection.send(ResponsesClientEvent.ofResponseCreate(request.build()));
var response = connection.finalResponse();
if (response.status().filter(ResponseStatus.COMPLETED::equals).isEmpty()) {
throw new IllegalStateException(
"Response ended with status " + response.status().orElse(null));
}
previousResponseId = response.id();
response.output().stream()
.flatMap(item -> item.message().stream())
.flatMap(message -> message.content().stream())
.flatMap(content -> content.outputText().stream())
.forEach(text -> System.out.println(text.text()));
}
} finally {
client.close();
}
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# Install: gem install openai async-websocket
# Set OPENAI_API_KEY in your environment.
require "async"
require "openai"
client = OpenAI::Client.new
request = {
model: "gpt-6-astra",
service_tier: :ultrafast
}
prompts = [
"Explain why the sky is blue in one sentence.",
"Now explain why sunsets look red."
]
Sync do |task|
task.with_timeout(120) do
client.responses.connect(request_options: { timeout: 10 }) do |connection|
prompts.each do |prompt|
connection.response.create(**request, input: prompt)
loop do
event = connection.receive
raise "Connection closed before the response finished" unless event
case event
when OpenAI::Responses::ResponseCompletedEvent
request[:previous_response_id] = event.response.id
puts(event.response.output_text)
break
when OpenAI::Responses::ResponseFailedEvent, OpenAI::Responses::ResponseIncompleteEvent,
OpenAI::Responses::ResponsesServerEvent::ResponseWsError
raise "Response failed: #{event.to_json}"
end
end
end
end
end
end
这些示例通过同一连接发送两个请求,并打印每个响应。Go、Java 和 Ruby 示例会等到收到完整响应后再打印。第二个请求会发送新的提示词,并将第一个响应的 ID 作为 previous_response_id 传入。在后续轮次中以及发送工具结果时,请继续使用同一连接。请参阅使用增量输入继续交互。
超快速模式也支持通过 SDK 发送 HTTP 请求。对于频繁调用工具的智能体应用,请使用持久 WebSocket 连接,以减少请求之间的网络开销。
1
2
3
4
5
6
7
8
9
10
11import OpenAI from "openai";
const client = new OpenAI();
const response = await client.responses.create({
model: "gpt-6-astra",
service_tier: "ultrafast",
input: "Explain why the sky is blue in one sentence.",
});
console.log(response.output_text);
1
2
3
4
5
6
7
8
9
10
11from openai import OpenAI
client = OpenAI()
response = client.responses.create(
model="gpt-6-astra",
input="Explain why the sky is blue in one sentence.",
service_tier="ultrafast",
)
print(response.output_text)
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()
response, err := client.Responses.New(context.Background(), responses.ResponseNewParams{
Model: "gpt-6-astra",
ServiceTier: responses.ResponseNewParamsServiceTierUltrafast,
Input: responses.ResponseNewParamsInputUnion{OfString: openai.String("Explain why the sky is blue in one sentence.")},
})
if err != nil {
panic(err)
}
fmt.Println(response.OutputText())
}
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.responses.ResponseCreateParams;
ResponseCreateParams params =
ResponseCreateParams.builder()
.model("gpt-6-astra")
.input("Explain why the sky is blue in one sentence.")
.serviceTier(ResponseCreateParams.ServiceTier.ULTRAFAST)
.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
11require "openai"
client = OpenAI::Client.new
response = client.responses.create(
model: "gpt-6-astra",
service_tier: :ultrafast,
input: "Explain why the sky is blue in one sentence."
)
puts(response.output_text)
1
2
3
4
5
6
7
8curl https://api.openai.com/v1/responses \
-H "Authorization: Bearer $OPENAI_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "gpt-6-astra",
"input": "Explain why the sky is blue in one sentence.",
"service_tier": "ultrafast"
}'
此示例会等待完整响应。如果希望边接收边显示输出,请启用流式传输。
请查看您所在组织对 GPT-6.2 Astra 和 GPT-6.2 Sol 的速率限制。
GPT-6.1 Sol 在超快速模式下的默认 Token 速率限制如下:
| API 使用层级 |
每分钟 Token 数(TPM) |
| Build |
1,000,000 |
| Launch |
4,000,000 |
| Grow |
40,000,000 |
GPT-6 Astra 在超快速模式下的默认 Token 速率限制如下:
| API 使用层级 |
每分钟 Token 数(TPM) |
| Build |
500,000 |
| Launch |
1,000,000 |
| Grow |
5,000,000 |
有关输入、缓存输入、缓存写入和输出的价格,请参阅超快速模式价格表。
GPT-6.2 Astra 和 GPT-6.2 Sol 的超快速模式仅支持美国数据驻留。有关其他服务层级的信息,请参阅数据驻留指南。
GPT-6.1 Sol 的超快速模式支持美国和欧盟的数据驻留,并支持全球处理。GPT-6 Astra 的超快速模式仅支持美国数据驻留和全球处理。