MiniMax-AI · NianJiuZst · Jul 14, 2026 · Jul 14, 2026
diff --git a/SDK.md b/SDK.md
@@ -118,7 +118,7 @@ const music = await sdk.music.generate({
 // Instrumental
 const instrumental = await sdk.music.generate({
   prompt: 'Cinematic orchestral',
-  instrumental: true,
+  is_instrumental: true,
 });
 
 // Auto-generate lyrics
@@ -135,7 +135,7 @@ const stream = await sdk.music.generate({
 });
 
 for await (const chunk of stream) {
-  // process audio chunks
+  // chunk contains decoded audio bytes
 }
 
 // Structured prompt

diff --git a/src/commands/music/cover.ts b/src/commands/music/cover.ts
@@ -11,15 +11,15 @@ import { pipeAudioStream } from '../../utils/audio-stream';
 import type { Config } from '../../config/schema';
 import type { GlobalFlags } from '../../types/flags';
 import type { CoverPreprocessRequest, CoverPreprocessResponse, MusicRequest, MusicResponse } from '../../types/api';
-import { musicCoverModel } from './models';
+import { MUSIC_COVER_MODELS, musicCoverModel } from './models';
 
 export default defineCommand({
   name: 'music cover',
-  description: 'Generate a cover version of a song based on reference audio (music-cover)',
+  description: 'Generate a cover version of a song based on reference audio',
   apiDocs: '/docs/api-reference/music-generation',
   usage: 'mmx music cover --prompt <text> (--audio <url> | --audio-file <path>) [--lyrics <text>] [--out <path>] [flags]',
   options: [
-    { flag: '--model <model>', description: 'Model: music-cover (default).' },
+    { flag: '--model <model>', description: 'Model: music-cover (default) or music-cover-free.' },
     { flag: '--prompt <text>', description: 'Target cover style, e.g. "Indie folk, acoustic guitar, warm male vocal"' },
     { flag: '--audio <url>', description: 'URL of the reference audio (mp3, wav, flac, etc. — 6s to 6min, max 50MB)' },
     { flag: '--audio-file <path>', description: 'Local reference audio file (auto base64-encoded)' },
@@ -37,6 +37,7 @@ export default defineCommand({
     'mmx music cover --prompt "Indie folk, acoustic guitar, warm male vocal" --audio https://example.com/song.mp3 --out cover.mp3',
     'mmx music cover --prompt "Jazz, piano, slow" --audio-file original.mp3 --lyrics-file lyrics.txt --out jazz_cover.mp3',
     'mmx music cover --prompt "Pop, upbeat" --audio https://example.com/ref.mp3 --seed 42 --out reproducible.mp3',
+    'mmx music cover --model music-cover-free --prompt "Jazz, piano, slow" --audio https://example.com/ref.mp3 --out free_cover.mp3',
   ],
   async run(config: Config, flags: GlobalFlags) {
     const prompt = flags.prompt as string | undefined;
@@ -72,10 +73,9 @@ export default defineCommand({
     const format = detectOutputFormat(config.output);
 
     const model = (flags.model as string) || musicCoverModel(config);
-    const VALID_MODELS = ['music-cover'];
-    if (flags.model && !VALID_MODELS.includes(model)) {
+    if (flags.model && !MUSIC_COVER_MODELS.includes(model as typeof MUSIC_COVER_MODELS[number])) {
       throw new CLIError(
-        `Invalid model "${model}". Valid models: ${VALID_MODELS.join(', ')}`,
+        `Invalid model "${model}". Valid models: ${MUSIC_COVER_MODELS.join(', ')}`,
         ExitCode.USAGE,
         'mmx music cover --model music-cover',
       );

diff --git a/src/commands/music/generate.ts b/src/commands/music/generate.ts
@@ -11,11 +11,11 @@ import { pipeAudioStream } from '../../utils/audio-stream';
 import type { Config } from '../../config/schema';
 import type { GlobalFlags } from '../../types/flags';
 import type { MusicRequest, MusicResponse } from '../../types/api';
-import { musicGenerateModel } from './models';
+import { MUSIC_GENERATE_MODELS, musicGenerateModel } from './models';
 
 export default defineCommand({
   name: 'music generate',
-  description: 'Generate a song (music-2.6 / music-2.5+ / music-2.5)',
+  description: 'Generate a song (music-2.6, including free tier)',
   apiDocs: '/docs/api-reference/music-generation',
   usage: 'mmx music generate --prompt <text> (--lyrics <text> | --instrumental | --lyrics-optimizer) [--out <path>] [flags]',
   options: [
@@ -36,7 +36,7 @@ export default defineCommand({
     { flag: '--structure <text>', description: 'Song structure, e.g. "verse-chorus-verse-bridge-chorus"' },
     { flag: '--references <text>', description: 'Reference tracks or artists, e.g. "similar to Ed Sheeran"' },
     { flag: '--extra <text>', description: 'Additional fine-grained requirements not covered above' },
-    { flag: '--model <model>', description: 'Model: music-2.6 (default), music-2.5+, or music-2.5.' },
+    { flag: '--model <model>', description: 'Model: music-2.6 (default), music-2.6-free, music-2.5+, or music-2.5.' },
     { flag: '--output-format <fmt>', description: 'Return format: hex (default, saved to file) or url (24h expiry, download promptly). When --stream, only hex.' },
     { flag: '--aigc-watermark', description: 'Embed AI-generated content watermark in audio for content provenance' },
     { flag: '--format <fmt>', description: `Audio format: ${formatList(MUSIC_FORMATS)} (default: mp3)` },
@@ -93,6 +93,14 @@ export default defineCommand({
       );
     }
 
+    if ((isInstrumental || lyricsOptimizer) && !prompt?.trim()) {
+      throw new CLIError(
+        '--prompt is required with --instrumental or --lyrics-optimizer.',
+        ExitCode.USAGE,
+        'mmx music generate --prompt <text> --instrumental',
+      );
+    }
+
     if (!isInstrumental && !lyricsOptimizer && !lyrics?.trim()) {
       throw new CLIError(
         'Lyrics are required. Add --lyrics or --lyrics-file, or use --instrumental for pure music, or --lyrics-optimizer to auto-generate.',
@@ -127,10 +135,9 @@ export default defineCommand({
     const outPath = (flags.out as string | undefined) ?? `music_${ts}.${ext}`;
 
     const model = (flags.model as string) || musicGenerateModel(config);
-    const VALID_MODELS = ['music-2.6', 'music-2.5+', 'music-2.5'];
-    if (flags.model && !VALID_MODELS.includes(model)) {
+    if (flags.model && !MUSIC_GENERATE_MODELS.includes(model as typeof MUSIC_GENERATE_MODELS[number])) {
       throw new CLIError(
-        `Invalid model "${model}". Valid models: ${VALID_MODELS.join(', ')}`,
+        `Invalid model "${model}". Valid models: ${MUSIC_GENERATE_MODELS.join(', ')}`,
         ExitCode.USAGE,
         'mmx music generate --model music-2.6',
       );

diff --git a/src/commands/music/models.ts b/src/commands/music/models.ts
@@ -1,13 +1,33 @@
 import type { Config } from '../../config/schema';
 
-export function musicGenerateModel(config: Config): string {
-  return config.defaultMusicModel ?? 'music-2.6';
+export const MUSIC_GENERATE_MODELS = [
+  'music-2.6',
+  'music-2.6-free',
+  'music-2.5+',
+  'music-2.5',
+] as const;
+
+export const MUSIC_COVER_MODELS = ['music-cover', 'music-cover-free'] as const;
+
+function includesModel(models: readonly string[], model: string): boolean {
+  return models.includes(model);
 }
 
-const VALID_COVER_MODELS = new Set(['music-cover']);
+export function musicGenerateModel(config: Config): string {
+  if (
+    config.defaultMusicModel
+    && includesModel(MUSIC_GENERATE_MODELS, config.defaultMusicModel)
+  ) {
+    return config.defaultMusicModel;
+  }
+  return 'music-2.6';
+}
 
 export function musicCoverModel(config: Config): string {
-  if (config.defaultMusicModel && VALID_COVER_MODELS.has(config.defaultMusicModel)) {
+  if (
+    config.defaultMusicModel
+    && includesModel(MUSIC_COVER_MODELS, config.defaultMusicModel)
+  ) {
     return config.defaultMusicModel;
   }
   return 'music-cover';

diff --git a/src/sdk/music/index.ts b/src/sdk/music/index.ts
@@ -7,7 +7,8 @@ import { ModelPartial } from "../types";
 import { SDKError } from "../../errors/base";
 import { ExitCode } from "../../errors/codes";
 import { toMerged } from "es-toolkit/object";
-import { musicGenerateModel } from "../../commands/music/models";
+import { MUSIC_GENERATE_MODELS, musicGenerateModel } from "../../commands/music/models";
+import { decodeAudioStream } from "../../utils/audio-stream";
 
 function hexToBuffer(hex: string): Buffer {
   if (!/^[0-9a-fA-F]*$/.test(hex)) {
@@ -64,20 +65,11 @@ export class MusicSDK extends Client {
       stream: true,
     });
 
-    const reader = res.body?.getReader();
-    if (!reader) {
+    if (!res.body) {
       throw new SDKError('No response body', ExitCode.GENERAL);
-    };
-
-    try {
-      while (true) {
-        const { done, value } = await reader.read();
-        if (done) break;
-        yield value;
-      }
-    } finally {
-      reader.releaseLock();
     }
+
+    yield* decodeAudioStream(res);
   }
 
   async generate(request: ModelPartial<MusicGenerateRequest> & { stream: true }): Promise<AsyncGenerator<Uint8Array<ArrayBuffer>>>;
@@ -140,16 +132,15 @@ export class MusicSDK extends Client {
     if (request.bpm)         structuredParts.push(`BPM: ${request.bpm as number}`);
     if (request.key)         structuredParts.push(`Key: ${request.key as string}`);
     if (request.avoid)       structuredParts.push(`Avoid: ${request.avoid as string}`);
-    if (request.useCase)     structuredParts.push(`Use case: ${request.useCase as string}`);
+    const useCase = request.useCase || request.use_case;
+    if (useCase)             structuredParts.push(`Use case: ${useCase}`);
     if (request.structure)   structuredParts.push(`Structure: ${request.structure as string}`);
     if (request.references)  structuredParts.push(`References: ${request.references as string}`);
     if (request.extra)       structuredParts.push(`Extra: ${request.extra as string}`);
 
-    let lyrics = request.lyrics;
     let prompt = request.prompt;
 
-    if (request.instrumental || !lyrics || lyrics === '无歌词' || lyrics === 'no lyrics') {
-      lyrics = '[intro] [outro]';
+    if (request.is_instrumental || request.instrumental) {
       structuredParts.push('Style: instrumental, no vocals, pure music');
     }
 
@@ -161,9 +152,43 @@ export class MusicSDK extends Client {
   }
 
   private validateParams(params: ModelPartial<MusicGenerateRequest>) {
+    const instrumental = params.instrumental;
+    const apiParams = { ...params };
+    const sdkOnlyFields: Array<keyof MusicGenerateRequest> = [
+      'instrumental',
+      'vocals',
+      'genre',
+      'mood',
+      'instruments',
+      'tempo',
+      'bpm',
+      'key',
+      'avoid',
+      'use_case',
+      'useCase',
+      'structure',
+      'references',
+      'extra',
+    ];
+    for (const field of sdkOnlyFields) delete apiParams[field];
+    if (
+      instrumental !== undefined
+      && params.is_instrumental !== undefined
+      && instrumental !== params.is_instrumental
+    ) {
+      throw new SDKError(
+        'instrumental and is_instrumental must not conflict',
+        ExitCode.USAGE,
+      );
+    }
+
+    const normalized: ModelPartial<MusicGenerateRequest> = {
+      ...apiParams,
+      is_instrumental: params.is_instrumental ?? instrumental,
+    };
     const {
       model, output_format, stream, prompt, lyrics, is_instrumental, lyrics_optimizer,
-    } = params;
+    } = normalized;
     if (is_instrumental && lyrics) {
       throw new SDKError('Cannot use is_instrumental with lyrics', ExitCode.USAGE);
     }
@@ -176,14 +201,20 @@ export class MusicSDK extends Client {
       throw new SDKError('At least one of prompt or lyrics or is_instrumental or lyrics_optimizer is required', ExitCode.USAGE);
     }
 
+    if ((is_instrumental || lyrics_optimizer) && !prompt?.trim()) {
+      throw new SDKError(
+        'prompt is required with is_instrumental or lyrics_optimizer',
+        ExitCode.USAGE,
+      );
+    }
+
     if (!is_instrumental && !lyrics_optimizer && !lyrics?.trim()) {
       throw new SDKError('lyrics is required', ExitCode.USAGE);
     }
 
-    const VALID_MODELS = ['music-2.6', 'music-2.5+', 'music-2.5'];
-    if (model && !VALID_MODELS.includes(model)) {
+    if (model && !MUSIC_GENERATE_MODELS.includes(model as typeof MUSIC_GENERATE_MODELS[number])) {
       throw new SDKError(
-        `Invalid model: ${model}. Valid models are ${VALID_MODELS.join(', ')}.`, 
+        `Invalid model: ${model}. Valid models are ${MUSIC_GENERATE_MODELS.join(', ')}.`,
         ExitCode.USAGE,
       );
     }
@@ -202,7 +233,10 @@ export class MusicSDK extends Client {
       );
     }
 
-    const targetPrompt = this.buildPrompt(params);
+    const targetPrompt = this.buildPrompt({
+      ...params,
+      is_instrumental,
+    });
 
     return toMerged({
       model: musicGenerateModel(this.config),
@@ -213,7 +247,7 @@ export class MusicSDK extends Client {
       },
       output_format: 'hex',
     }, {
-      ...params,
+      ...normalized,
       prompt: targetPrompt,
     });
   }

diff --git a/src/utils/audio-stream.ts b/src/utils/audio-stream.ts
@@ -4,12 +4,9 @@ interface SseAudioPayload {
   data?: { audio?: string; status?: number };
 }
 
-export async function pipeAudioStream(response: Response): Promise<void> {
-  process.stdout.on('error', (err: NodeJS.ErrnoException) => {
-    if (err.code === 'EPIPE') process.exit(0);
-    throw err;
-  });
-
+export async function* decodeAudioStream(
+  response: Response,
+): AsyncGenerator<Uint8Array<ArrayBuffer>> {
   for await (const event of parseSSE(response)) {
     if (!event.data || event.data === '[DONE]') break;
 
@@ -21,7 +18,17 @@ export async function pipeAudioStream(response: Response): Promise<void> {
     const hex = parsed.data?.audio;
     if (!hex) continue;
 
-    const chunk = Buffer.from(hex, 'hex');
+    yield Uint8Array.from(Buffer.from(hex, 'hex'));
+  }
+}
+
+export async function pipeAudioStream(response: Response): Promise<void> {
+  process.stdout.on('error', (err: NodeJS.ErrnoException) => {
+    if (err.code === 'EPIPE') process.exit(0);
+    throw err;
+  });
+
+  for await (const chunk of decodeAudioStream(response)) {
     if (!process.stdout.write(chunk)) {
       await new Promise<void>(r => process.stdout.once('drain', r));
     }

diff --git a/test/commands/music/cover.test.ts b/test/commands/music/cover.test.ts
@@ -151,6 +151,29 @@ describe('music cover command', () => {
     ).rejects.toThrow('Invalid model');
   });
 
+  it('accepts the official music-cover-free model', async () => {
+    let captured = '';
+    const origLog = console.log;
+    console.log = (msg: string) => { captured += msg; };
+
+    try {
+      await coverCommand.execute(
+        { ...baseConfig, dryRun: true, output: 'json' as const },
+        {
+          ...baseFlags,
+          dryRun: true,
+          prompt: 'Folk cover',
+          audio: 'https://example.com/ref.mp3',
+          model: 'music-cover-free',
+        },
+      );
+    } finally {
+      console.log = origLog;
+    }
+
+    expect(JSON.parse(captured).request.model).toBe('music-cover-free');
+  });
+
   it('rejects invalid audio format', async () => {
     await expect(
       coverCommand.execute(

diff --git a/test/commands/music/generate.test.ts b/test/commands/music/generate.test.ts
@@ -181,6 +181,38 @@ describe('music generate command', () => {
     expect(parsed.request.model).toBe('music-2.5');
   });
 
+  it('accepts the official music-2.6-free model', async () => {
+    let captured = '';
+    const origLog = console.log;
+    console.log = (msg: string) => { captured += msg; };
+
+    try {
+      await generateCommand.execute(
+        { ...baseConfig, dryRun: true, output: 'json' as const },
+        {
+          ...baseFlags,
+          dryRun: true,
+          prompt: 'Folk',
+          lyrics: 'la la',
+          model: 'music-2.6-free',
+        },
+      );
+    } finally {
+      console.log = origLog;
+    }
+
+    expect(JSON.parse(captured).request.model).toBe('music-2.6-free');
+  });
+
+  it('requires prompt for instrumental generation', async () => {
+    await expect(
+      generateCommand.execute(
+        { ...baseConfig, dryRun: true },
+        { ...baseFlags, dryRun: true, instrumental: true },
+      ),
+    ).rejects.toThrow('--prompt is required with --instrumental');
+  });
+
   it('rejects invalid audio format', async () => {
     await expect(
       generateCommand.execute(