MiniMax Music 2.6 - MiniMax's new AI music generation model

What is MiniMax Music 2.6?

MiniMax Music 2.6 is a brand-new AI music generation model launched by MiniMax, achieving a comprehensive evolution from the underlying architecture to the creative experience. This version compresses the initial package latency to within 20 seconds, fundamentally solving the "blind box" pain point, and achieving precise BPM locking, fine-grained segment structure control, and deep emotional expression. In terms of acoustic quality, vocals are more natural and realistic, low-frequency performance is significantly enhanced, and the timbre of traditional Chinese instruments is accurately reproduced. It also introduces a cross-style cover function and the open-source AI Agent Music Skill ecosystem, supporting fun features such as melody transfer to pet singers.

Main features of MiniMax Music 2.6

• Intelligent music generation Input natural language descriptions or lyrics, and AI will automatically generate a complete song. It supports multiple genres (pop, rock, electronic, Chinese style, etc.) and can precisely control BPM, key, and 14 paragraph structures (such as [Intro], [Verse], [Chorus], [Bridge], etc.).
• Covers and Remakes After uploading reference audio, the model can extract the melody skeleton for cross-style adaptation (such as changing pop to jazz or folk to heavy metal), or retain the original melody and embed new lyrics to generate cover songs, thus realizing secondary music creation.
• AI Agent Music Skill (Open Source Ecosystem) Three open-source skills for developers:
  • minimax-music-gen2The agent automatically generates music based on natural language descriptions.
  • minimax-music-playlistAutomatically create personalized playlists based on user tastes and scenarios.
  • buddy-singsAllow virtual characters to improvise and sing in the first person, or convert pet voices into songs.
• Professional-grade audio control It supports precise tempo locking, tonality control, and fine-grained paragraph structure planning, solving the uncontrollable problem of "opening blind boxes" in AI music and ensuring that creative intentions are faithfully reproduced.
• Multilingual voice synthesis Optimizes vocal performance for a more natural and realistic singing experience; supports lyric generation in multiple languages including Chinese and English; and accurately simulates the dynamic changes of erhu, flute, guzheng, and opera singing styles in traditional Chinese music scenes.

How to use MiniMax Music 2.6

• Visit the official website and log inVisit the official website of Music 2.6: https://www.minimaxi.com/audio/music and log in to your account (new users need to register). It is currently in a 14-day free beta testing period, and C-end users can enjoy 500 free creation quotas per day.
• Select creation mode On the main interface, select the "Music Generation" mode to start creating. If you need to adapt an existing song, you can select the "Cover" mode.
• Input creation prompts In the input box, describe the music style, mood, scene, or purpose you want (e.g., "lyrical pop, Chinese song about heartbreak, slow tempo"), or you can directly paste the lyrics.
• Set music parameters(Optional): Set the BPM (beat speed) and key (such as C major or A minor) as needed, and use structure tags to define song sections (such as [Intro][Verse][Chorus][Outro]) to control the music structure.
• Generate and Preview Click the generate button, and you will get the preliminary audio in about 20 seconds. You can preview the effect online.
• Refinement and Adjustment If the effect is not satisfactory, the prompt words or parameters can be adjusted and regenerated, supporting multiple iterations for optimization.
• Download and share Once you are satisfied, you can download the high-quality audio file or share it directly to social media platforms.
• Using the Cover feature(Advanced): If you need to cover a song or change the style, upload a reference audio file, select the target style or fill in new lyrics to generate an adapted version.
• Access API(Developers): Developers can obtain API keys through the MiniMax Open Platform and integrate the Music 2.6 interface into their own applications. During the internal testing period, 100 free calls will be given every day.

MiniMax Music 2.6's core advantages

• Lightning-fast generation experience After a deep restructuring of the underlying architecture, the latency of the first packet was compressed to [a specific value].Within 20 secondsCreators can get initial audio feedback after entering a prompt word, requiring only "the time it takes to take a deep breath," eliminating long waits.
• Breakthrough in Precision and Controllability Completely solve the pain point of "opening blind boxes" in AI music, and achieve...BPM and tone precisely locked,14 types of paragraph structure fine control(such as [Verse], [Chorus], [Bridge], etc.), and deep emotional control to ensure that the creative intent is faithfully reproduced.
• Acoustic quality fully upgraded
  • naturalization of human voiceShedding the mechanical singing style, it presents a more casual and free sense of authentic singing.
  • Low frequency reinforcementThe mid-low frequencies extend deeper and engage more tightly, resolving the muddiness issue in electronic music and game soundtracks.
  • Traditional Chinese style optimizationPrecisely simulates the vibrato of the erhu, the breath pauses of the flute, the strumming of the guzheng, and the dynamic changes of opera singing.
• Industry-first Cover feature It supports cross-style migration (such as pop to jazz, folk to heavy metal) and intelligent lyric writing for covers, providing new possibilities for music re-creation.
• AI Agent Ecosystem Integration Open sourceminimax-music-gen2,minimax-music-playlist,buddy-singsThree major Music Skills enable the AI Agent to generate music, arrange playlists, and perform character singing.
• Deep optimization for Chinese scenarios : Special optimizations have been made for Chinese pop music to better suit the aesthetic habits of Chinese users in terms of melody lines, lyric pronunciation, and emotional expression.

MiniMax Music 2.6 official website address

• Official website addresshttps://www.minimaxi.com/audio/music

Key information and usage requirements for MiniMax Music 2.6

• Release timeApril 10, 2026
• CompanyMiniMax (Rare Universe Technology)
• Product PositioningThe next-generation AI music generation model is based on the abab 7 series large model architecture.
• Version FeaturesA complete evolution from the underlying engine to creation tools.
• Generation speedFirst packet delay <20 seconds
• Structural controlSupports 14 paragraph tags ([Intro], [Verse], [Chorus], [Bridge], [Hook], etc.)
• Music DimensionPrecise control of BPM (beat speed) and tonality (C major, A minor, etc.)
• Tonal performanceNatural vocals, enhanced low frequencies, and optimized for traditional Chinese musical instruments.
• Cover functionCross-style transfer and intelligent lyric writing for cover songs
• AI Agent SkillThe three main open-source technologies are minimax-music-gen2, minimax-music-playlist, and buddy-sings.
• Real-time generationReturn preliminary audio results within 20 seconds.
• Internal testing period14-day free beta testing period (starting April 10, 2026)
• C-end users500 free creation quotas per day
• DevelopersAn additional 100 free API calls per day will be provided.
• Exceeding the limitPlease pay attention to the official payment policy after the closed beta period ends.
• Prompt wordsSupports natural language descriptions (style, mood, scene) or direct pasting of lyrics.
• Structural tagsIt is recommended to use tags such as [Verse] and [Chorus] to clearly define the paragraph structure.
• Parameter settingsYou can specify the BPM range and tone to improve generation accuracy.

Analysis of competing products similar to MiniMax Music 2.6

Application scenarios of MiniMax Music 2.6

• Personal music creation Ordinary users do not need any musical background; they can create their own original songs using natural language descriptions. These songs are suitable for birthday wishes, emotional expressions, and commemorative gifts, such as remaking songs from the past for family members as special gifts.
• Games and Interactive Entertainment It is specifically optimized for low-to-mid frequency performance in game soundtracks, and can generate background music, battle BGM, and scene sound effects that match the game atmosphere. Developers can use the API to generate audio resources adapted to different level styles in batches.
• Short videos and social media content It provides customized background music for short video creators, podcast hosts, and live streamers, supports the rapid generation of background music that matches the mood of the content, and solves the problem of restricted copyrighted music.
• AI Agent and Intelligent Assistant Integration Through open-source Music Skills (minimax-music-gen2, minimax-music-playlist, buddy-sings), AI Agent is equipped with capabilities such as music generation, personalized playlist arrangement, and virtual character singing, empowering applications such as intelligent customer service and virtual companionship.
• Chinese style and traditional cultural content Using specialized optimization capabilities for traditional Chinese style, we can generate musical works that accurately simulate the erhu, flute, guzheng, and opera singing styles for scenarios such as ancient style videos, Hanfu displays, traditional culture promotion, and opera education.