跳到主要内容

OpenAI 标志OpenAI

通过 Mastra 的模型路由器访问 47 个 OpenAI 模型。系统会自动使用 OPENAI_API_KEY 环境变量完成身份验证。

更多信息请参阅OpenAI 文档

.env
OPENAI_API_KEY=your-api-key
src/mastra/agents/my-agent.ts
import { Agent } from '@mastra/core/agent'

const agent = new Agent({
id: 'my-agent',
name: 'My Agent',
instructions: 'You are a helpful assistant',
model: 'openai/chatgpt-image-latest',
})

// Generate a response
const response = await agent.generate('Hello!')

// Stream a response
const stream = await agent.stream('Tell me a story')
for await (const chunk of stream) {
console.log(chunk)
}

模型
模型的直接链接

ModelContextToolsReasoningImageAudioVideoInput $/1MOutput $/1M
openai/chatgpt-image-latest
openai/gpt-4.11.0M$2$8
openai/gpt-4.1-mini1.0M$0.40$2
openai/gpt-4o128K$3$10
openai/gpt-4o-2024-08-06128K$3$10
openai/gpt-4o-2024-11-20128K$3$10
openai/gpt-4o-mini128K$0.15$0.60
openai/gpt-5400K$1$10
openai/gpt-5-mini400K$0.25$2
openai/gpt-5-nano400K$0.05$0.40
openai/gpt-5-pro400K$15$120
openai/gpt-5.1400K$1$10
openai/gpt-5.2400K$2$14
openai/gpt-5.2-chat-latest128K$2$14
openai/gpt-5.2-pro400K$21$168
openai/gpt-5.3-chat-latest128K$2$14
openai/gpt-5.3-codex400K$2$14
openai/gpt-5.3-codex-spark128K$2$14
openai/gpt-5.41.1M$3$15
openai/gpt-5.4-mini400K$0.75$5
openai/gpt-5.4-nano400K$0.20$1
openai/gpt-5.4-pro1.1M$30$180
openai/gpt-5.51.1M$5$30
openai/gpt-5.5-pro1.1M$30$180
openai/gpt-5.61.1M$5$30
openai/gpt-5.6-luna1.1M$0.20$1
openai/gpt-5.6-sol1.1M$5$30
openai/gpt-5.6-terra1.1M$2$12
openai/gpt-image-1-mini
openai/gpt-image-1.5
openai/gpt-image-2$5$30
openai/gpt-realtime-2.1128K$4$24
openai/o3200K$2$8
openai/o3-pro200K$20$80
openai/text-embedding-3-large8K$0.13
openai/text-embedding-3-small8K$0.02
openai/text-embedding-ada-0028K$0.10
37 available models

高级配置
高级配置的直接链接

自定义请求头
自定义请求头的直接链接

src/mastra/agents/my-agent.ts
const agent = new Agent({
id: 'custom-agent',
name: 'custom-agent',
model: {
id: 'openai/chatgpt-image-latest',
apiKey: process.env.OPENAI_API_KEY,
headers: {
'X-Custom-Header': 'value',
},
},
})

动态选择模型
动态选择模型的直接链接

src/mastra/agents/my-agent.ts
const agent = new Agent({
id: 'dynamic-agent',
name: 'Dynamic Agent',
model: ({ requestContext }) => {
const useAdvanced = requestContext.task === 'complex'
return useAdvanced ? 'openai/text-embedding-ada-002' : 'openai/chatgpt-image-latest'
},
})

Provider 选项
Provider 选项的直接链接

OpenAI 通过 providerOptions 参数支持以下 Provider 特有选项:

const response = await agent.generate('Hello!', {
providerOptions: {
openai: {
// See available options in the table below
},
},
})

可用选项
可用选项的直接链接

conversation?:

string | null | undefined

include?:

("web_search_call.results" | "file_search_call.results" | "message.output_text.logprobs" | "reasoning.encrypted_content")[] | null | undefined

instructions?:

string | null | undefined

logprobs?:

number | boolean | undefined

maxToolCalls?:

number | null | undefined

metadata?:

any

parallelToolCalls?:

boolean | null | undefined

previousResponseId?:

string | null | undefined

promptCacheKey?:

string | null | undefined

promptCacheOptions?:

{ mode?: "explicit" | "implicit" | undefined; ttl?: "30m" | undefined; } | undefined

promptCacheRetention?:

"in_memory" | "24h" | null | undefined

reasoningEffort?:

string | null | undefined

reasoningMode?:

"standard" | "pro" | undefined

reasoningContext?:

"auto" | "current_turn" | "all_turns" | undefined

reasoningSummary?:

string | null | undefined

safetyIdentifier?:

string | null | undefined

serviceTier?:

"auto" | "default" | "flex" | "priority" | null | undefined

store?:

boolean | null | undefined
Controls whether OpenAI stores your API requests for model training. Required to be "false" if your organization has zero data retention enabled. See: https://platform.openai.com/docs/guides/your-data#zero-data-retention

passThroughUnsupportedFiles?:

boolean | undefined

strictJsonSchema?:

boolean | null | undefined

textVerbosity?:

"low" | "medium" | "high" | null | undefined

truncation?:

"auto" | "disabled" | null | undefined

user?:

string | null | undefined

systemMessageMode?:

"remove" | "system" | "developer" | undefined

forceReasoning?:

boolean | undefined

allowedTools?:

{ toolNames: string[]; mode?: "auto" | "required" | undefined; } | undefined

直接安装 Provider
直接安装 Provider的直接链接

你也可以直接将此 Provider 安装为独立 package,用它代替 Mastra 模型路由字符串。详情请参阅package 文档

npm install @ai-sdk/openai

有关此 Provider 的详细说明,请参阅 AI SDK OpenAI Provider 文档