feat: add MiniMax as OpenAI-compatible LLM provider

README.md

@@ -58,7 +58,23 @@ const text = await generateText({
 });
 ```
 
-Providers: [OpenAI](https://modelfusion.dev/integration/model-provider/openai), [OpenAI compatible](https://modelfusion.dev/integration/model-provider/openaicompatible), [Llama.cpp](https://modelfusion.dev/integration/model-provider/llamacpp), [Ollama](https://modelfusion.dev/integration/model-provider/ollama), [Mistral](https://modelfusion.dev/integration/model-provider/mistral), [Hugging Face](https://modelfusion.dev/integration/model-provider/huggingface), [Cohere](https://modelfusion.dev/integration/model-provider/cohere)
+#### generateText with MiniMax
+
+```ts
+import { generateText, openaicompatible } from "modelfusion";
+
+const text = await generateText({
+  model: openaicompatible
+    .ChatTextGenerator({
+      api: openaicompatible.MiniMaxApi(),
+      model: "MiniMax-M2.5",
+    })
+    .withTextPrompt(),
+  prompt: "Write a short story about a robot learning to love:",
+});
+```
+
+Providers: [OpenAI](https://modelfusion.dev/integration/model-provider/openai), [OpenAI compatible](https://modelfusion.dev/integration/model-provider/openaicompatible), [Llama.cpp](https://modelfusion.dev/integration/model-provider/llamacpp), [Ollama](https://modelfusion.dev/integration/model-provider/ollama), [Mistral](https://modelfusion.dev/integration/model-provider/mistral), [Hugging Face](https://modelfusion.dev/integration/model-provider/huggingface), [Cohere](https://modelfusion.dev/integration/model-provider/cohere), [MiniMax](https://platform.minimaxi.com/document/Fast%20access)
 
 #### streamText
 
@@ -305,7 +321,7 @@ const embeddings = await embedMany({
 });
 ```
 
-Providers: [OpenAI](https://modelfusion.dev/integration/model-provider/openai), [OpenAI compatible](https://modelfusion.dev/integration/model-provider/openaicompatible), [Llama.cpp](https://modelfusion.dev/integration/model-provider/llamacpp), [Ollama](https://modelfusion.dev/integration/model-provider/ollama), [Mistral](https://modelfusion.dev/integration/model-provider/mistral), [Hugging Face](https://modelfusion.dev/integration/model-provider/huggingface), [Cohere](https://modelfusion.dev/integration/model-provider/cohere)
+Providers: [OpenAI](https://modelfusion.dev/integration/model-provider/openai), [OpenAI compatible](https://modelfusion.dev/integration/model-provider/openaicompatible), [Llama.cpp](https://modelfusion.dev/integration/model-provider/llamacpp), [Ollama](https://modelfusion.dev/integration/model-provider/ollama), [Mistral](https://modelfusion.dev/integration/model-provider/mistral), [Hugging Face](https://modelfusion.dev/integration/model-provider/huggingface), [Cohere](https://modelfusion.dev/integration/model-provider/cohere), [MiniMax](https://platform.minimaxi.com/document/Fast%20access)
 
 ### [Classify Value](https://modelfusion.dev/guide/function/classify)
 

packages/modelfusion/src/model-provider/openai-compatible/MiniMaxApiConfiguration.integration.test.ts

@@ -0,0 +1,60 @@
+/**
+ * Integration tests for MiniMax API via OpenAI-compatible interface.
+ *
+ * These tests require a valid MINIMAX_API_KEY environment variable.
+ * Run with: MINIMAX_API_KEY=your_key npx vitest run src/model-provider/openai-compatible/MiniMaxApiConfiguration.integration.test.ts
+ */
+import { describe, it, expect } from "vitest";
+import { generateText } from "../../model-function/generate-text/generateText";
+import { streamText } from "../../model-function/generate-text/streamText";
+import { MiniMaxApiConfiguration } from "./MiniMaxApiConfiguration";
+import { OpenAICompatibleChatModel } from "./OpenAICompatibleChatModel";
+import { arrayFromAsync } from "../../test/arrayFromAsync";
+
+const SKIP = !process.env.MINIMAX_API_KEY;
+
+describe.skipIf(SKIP)("MiniMax integration", () => {
+  it("should generate text with MiniMax-M2.5", async () => {
+    const text = await generateText({
+      model: new OpenAICompatibleChatModel({
+        api: new MiniMaxApiConfiguration(),
+        model: "MiniMax-M2.5",
+        maxGenerationTokens: 50,
+      }).withTextPrompt(),
+      prompt: "Say hello in one sentence.",
+    });
+
+    expect(text).toBeTruthy();
+    expect(text.length).toBeGreaterThan(0);
+  }, 30_000);
+
+  it("should stream text with MiniMax-M2.5", async () => {
+    const stream = await streamText({
+      model: new OpenAICompatibleChatModel({
+        api: new MiniMaxApiConfiguration(),
+        model: "MiniMax-M2.5",
+        maxGenerationTokens: 50,
+      }).withTextPrompt(),
+      prompt: "Count from 1 to 5.",
+    });
+
+    const chunks = await arrayFromAsync(stream);
+    expect(chunks.length).toBeGreaterThan(0);
+    const fullText = chunks.join("");
+    expect(fullText).toBeTruthy();
+  }, 60_000);
+
+  it("should generate text with MiniMax-M2.7", async () => {
+    const text = await generateText({
+      model: new OpenAICompatibleChatModel({
+        api: new MiniMaxApiConfiguration(),
+        model: "MiniMax-M2.7",
+        maxGenerationTokens: 50,
+      }).withTextPrompt(),
+      prompt: "What is 2+2? Answer with just the number.",
+    });
+
+    expect(text).toBeTruthy();
+    expect(text).toContain("4");
+  }, 30_000);
+});

packages/modelfusion/src/model-provider/openai-compatible/MiniMaxApiConfiguration.test.ts

@@ -0,0 +1,198 @@
+import { streamText } from "../../model-function/generate-text/streamText";
+import { generateText } from "../../model-function/generate-text/generateText";
+import { StreamingTestServer } from "../../test/StreamingTestServer";
+import { JsonTestServer } from "../../test/JsonTestServer";
+import { arrayFromAsync } from "../../test/arrayFromAsync";
+import { MiniMaxApiConfiguration } from "./MiniMaxApiConfiguration";
+import { OpenAICompatibleChatModel } from "./OpenAICompatibleChatModel";
+
+describe("MiniMaxApiConfiguration", () => {
+  it("should set the correct base URL", () => {
+    const api = new MiniMaxApiConfiguration({ apiKey: "test-key" });
+    expect(api.assembleUrl("/chat/completions")).toBe(
+      "https://api.minimax.io:443/v1/chat/completions"
+    );
+  });
+
+  it("should set the correct provider name", () => {
+    const api = new MiniMaxApiConfiguration({ apiKey: "test-key" });
+    expect(api.provider).toBe("openaicompatible-minimax");
+  });
+
+  it("should allow custom base URL overrides", () => {
+    const api = new MiniMaxApiConfiguration({
+      apiKey: "test-key",
+      baseUrl: { host: "custom.minimax.io" },
+    });
+    expect(api.assembleUrl("/chat/completions")).toBe(
+      "https://custom.minimax.io:443/v1/chat/completions"
+    );
+  });
+});
+
+describe("MiniMax streamText", () => {
+  const server = new StreamingTestServer(
+    "https://api.minimax.io/v1/chat/completions"
+  );
+
+  server.setupTestEnvironment();
+
+  describe("simple hello world stream", () => {
+    beforeEach(() => {
+      server.responseChunks = [
+        `data: {"id":"chatcmpl-minimax-001","object":"chat.completion.chunk","created":1703439030,"model":"MiniMax-M2.5",` +
+          `"choices":[{"index":0,"delta":{"role":"assistant","content":""},"finish_reason":null}]}\n\n`,
+        `data: {"id":"chatcmpl-minimax-001","object":"chat.completion.chunk","created":1703439030,"model":"MiniMax-M2.5",` +
+          `"choices":[{"index":0,"delta":{"content":"Hello"},"finish_reason":null}]}\n\n`,
+        `data: {"id":"chatcmpl-minimax-001","object":"chat.completion.chunk","created":1703439030,"model":"MiniMax-M2.5",` +
+          `"choices":[{"index":0,"delta":{"content":", "},"finish_reason":null}]}\n\n`,
+        `data: {"id":"chatcmpl-minimax-001","object":"chat.completion.chunk","created":1703439030,"model":"MiniMax-M2.5",` +
+          `"choices":[{"index":0,"delta":{"content":"world!"},"finish_reason":null}]}\n\n`,
+        `data: {"id":"chatcmpl-minimax-001","object":"chat.completion.chunk","created":1703439030,"model":"MiniMax-M2.5",` +
+          `"choices":[{"index":0,"delta":{},"finish_reason":"stop"}]}\n\n`,
+        `data: [DONE]\n\n`,
+      ];
+    });
+
+    it("should return a text stream", async () => {
+      const stream = await streamText({
+        model: new OpenAICompatibleChatModel({
+          api: new MiniMaxApiConfiguration({ apiKey: "test-key" }),
+          model: "MiniMax-M2.5",
+        }).withTextPrompt(),
+        prompt: "hello",
+      });
+
+      // note: space moved to next chunk due to trimming
+      expect(await arrayFromAsync(stream)).toStrictEqual([
+        "Hello",
+        ",",
+        " world!",
+      ]);
+    });
+
+    it("should return full text", async () => {
+      const { textPromise } = await streamText({
+        model: new OpenAICompatibleChatModel({
+          api: new MiniMaxApiConfiguration({ apiKey: "test-key" }),
+          model: "MiniMax-M2.5",
+        }).withTextPrompt(),
+        prompt: "hello",
+        fullResponse: true,
+      });
+
+      expect(await textPromise).toStrictEqual("Hello, world!");
+    });
+  });
+});
+
+describe("MiniMax generateText", () => {
+  const server = new JsonTestServer(
+    "https://api.minimax.io/v1/chat/completions"
+  );
+
+  server.setupTestEnvironment();
+
+  it("should return generated text", async () => {
+    server.responseBodyJson = {
+      id: "chatcmpl-minimax-002",
+      object: "chat.completion",
+      created: 1703439030,
+      model: "MiniMax-M2.5",
+      choices: [
+        {
+          index: 0,
+          message: {
+            role: "assistant",
+            content: "Hello, world!",
+          },
+          finish_reason: "stop",
+        },
+      ],
+      usage: {
+        prompt_tokens: 10,
+        completion_tokens: 5,
+        total_tokens: 15,
+      },
+    };
+
+    const text = await generateText({
+      model: new OpenAICompatibleChatModel({
+        api: new MiniMaxApiConfiguration({ apiKey: "test-key" }),
+        model: "MiniMax-M2.5",
+      }).withTextPrompt(),
+      prompt: "hello",
+    });
+
+    expect(text).toStrictEqual("Hello, world!");
+  });
+
+  it("should work with MiniMax-M2.7 model", async () => {
+    server.responseBodyJson = {
+      id: "chatcmpl-minimax-003",
+      object: "chat.completion",
+      created: 1703439030,
+      model: "MiniMax-M2.7",
+      choices: [
+        {
+          index: 0,
+          message: {
+            role: "assistant",
+            content: "I am MiniMax M2.7.",
+          },
+          finish_reason: "stop",
+        },
+      ],
+      usage: {
+        prompt_tokens: 8,
+        completion_tokens: 6,
+        total_tokens: 14,
+      },
+    };
+
+    const text = await generateText({
+      model: new OpenAICompatibleChatModel({
+        api: new MiniMaxApiConfiguration({ apiKey: "test-key" }),
+        model: "MiniMax-M2.7",
+      }).withTextPrompt(),
+      prompt: "What model are you?",
+    });
+
+    expect(text).toStrictEqual("I am MiniMax M2.7.");
+  });
+
+  it("should work with temperature setting", async () => {
+    server.responseBodyJson = {
+      id: "chatcmpl-minimax-004",
+      object: "chat.completion",
+      created: 1703439030,
+      model: "MiniMax-M2.5",
+      choices: [
+        {
+          index: 0,
+          message: {
+            role: "assistant",
+            content: "Creative response here.",
+          },
+          finish_reason: "stop",
+        },
+      ],
+      usage: {
+        prompt_tokens: 12,
+        completion_tokens: 4,
+        total_tokens: 16,
+      },
+    };
+
+    const text = await generateText({
+      model: new OpenAICompatibleChatModel({
+        api: new MiniMaxApiConfiguration({ apiKey: "test-key" }),
+        model: "MiniMax-M2.5",
+        temperature: 0.7,
+      }).withTextPrompt(),
+      prompt: "Be creative",
+    });
+
+    expect(text).toStrictEqual("Creative response here.");
+  });
+});

packages/modelfusion/src/model-provider/openai-compatible/MiniMaxApiConfiguration.ts

@@ -0,0 +1,43 @@
+import {
+  BaseUrlApiConfigurationWithDefaults,
+  PartialBaseUrlPartsApiConfigurationOptions,
+} from "../../core/api/BaseUrlApiConfiguration";
+import { loadApiKey } from "../../core/api/loadApiKey";
+import { OpenAICompatibleApiConfiguration } from "./OpenAICompatibleApiConfiguration";
+
+/**
+ * Configuration for the MiniMax API.
+ *
+ * It calls the API at https://api.minimax.io/v1 and uses the `MINIMAX_API_KEY` api key environment variable.
+ *
+ * @see https://platform.minimaxi.com/document/Fast%20access
+ */
+export class MiniMaxApiConfiguration
+  extends BaseUrlApiConfigurationWithDefaults
+  implements OpenAICompatibleApiConfiguration
+{
+  constructor(
+    settings: PartialBaseUrlPartsApiConfigurationOptions & {
+      apiKey?: string;
+    } = {}
+  ) {
+    super({
+      ...settings,
+      headers: {
+        Authorization: `Bearer ${loadApiKey({
+          apiKey: settings.apiKey,
+          environmentVariableName: "MINIMAX_API_KEY",
+          description: "MiniMax",
+        })}`,
+      },
+      baseUrlDefaults: {
+        protocol: "https",
+        host: "api.minimax.io",
+        port: "443",
+        path: "/v1",
+      },
+    });
+  }
+
+  readonly provider = "openaicompatible-minimax" as const;
+}

packages/modelfusion/src/model-provider/openai-compatible/OpenAICompatibleFacade.ts

@@ -1,5 +1,6 @@
 import { PartialBaseUrlPartsApiConfigurationOptions } from "../../core/api/BaseUrlApiConfiguration";
 import { FireworksAIApiConfiguration } from "./FireworksAIApiConfiguration";
+import { MiniMaxApiConfiguration } from "./MiniMaxApiConfiguration";
 import {
   OpenAICompatibleChatModel,
   OpenAICompatibleChatSettings,
@@ -27,6 +28,21 @@ export function FireworksAIApi(
   return new FireworksAIApiConfiguration(settings);
 }
 
+/**
+ * Configuration for the MiniMax API.
+ *
+ * It calls the API at https://api.minimax.io/v1 and uses the `MINIMAX_API_KEY` api key environment variable.
+ *
+ * @see https://platform.minimaxi.com/document/Fast%20access
+ */
+export function MiniMaxApi(
+  settings: PartialBaseUrlPartsApiConfigurationOptions & {
+    apiKey?: string;
+  } = {}
+) {
+  return new MiniMaxApiConfiguration(settings);
+}
+
 /**
  * Configuration for the Perplexity API.
  *

packages/modelfusion/src/model-provider/openai-compatible/index.ts

@@ -1,4 +1,5 @@
 export * from "./FireworksAIApiConfiguration";
+export * from "./MiniMaxApiConfiguration";
 export * from "./OpenAICompatibleApiConfiguration";
 export * from "./OpenAICompatibleChatModel";
 export * from "./OpenAICompatibleCompletionModel";