MiniMax Music 2.6 is a text-to-audio model that generates full musical tracks by combining style descriptions, lyrics, and structural instructions. Instead of assembling fragments, it creates end-to-end compositions with intros, verses, choruses, and transitions that follow a defined musical arc.
One of the defining changes in this version is the shift toward intent-aware composition. The model doesn’t just interpret genre tags, it understands how a track should evolve over time, including buildup, tension, and release.
Unlike earlier AI music systems that output short loops, Music 2.6 generates complete songs with vocals and instrumentals. A single prompt can result in a track that includes structure, arrangement, and performance dynamics.
Creators can define the composition using tags such as [Verse], [Chorus], [Bridge], or [Outro]. This allows precise control over how the song unfolds, making the output predictable and editable at a structural level.
A major addition in this version is the Cover feature, which extracts the melodic core of an existing track and reinterprets it in a new style. This enables transformations such as turning a folk melody into an electronic track while preserving recognizability.
Music 2.6 is designed to follow temporal progression in music, not just static style prompts. It can start with minimal instrumentation, gradually introduce layers, and build toward a defined climax—mirroring real-world composition techniques.
Instead of relying on presets, users can describe how a track should evolve. For example, prompts can specify emotional transitions or energy curves, and the model will reflect that progression in the final output.
The model introduces more natural handling of vocals and melody, including subtle imperfections that make tracks feel less mechanical and more human.
One of the key technical upgrades is improved handling of bass and low-end frequencies, resulting in tighter drums and clearer sub-bass. This is especially noticeable in genres like electronic, hip-hop, and cinematic scoring.
Music can be generated at up to 44.1kHz sample rate and 256kbps bitrate, making it suitable for real-world use rather than just prototyping.
The model adapts to a wide range of styles, from ambient and lo-fi to orchestral and high-energy electronic, while maintaining coherence within each genre.
Music 2.6 significantly reduces the time between prompt submission and initial output, allowing creators to iterate quickly without long waiting cycles.
Users can define parameters such as tempo, key, structure, and emotional tone directly in the prompt. The model follows these instructions with improved accuracy compared to earlier versions.
You can either provide full lyrics for structured songs or rely on the model to generate lyrics automatically based on a theme or concept.
| Capability | How It Works | Why It Matters |
|---|---|---|
| Structure tags | Define sections inside lyrics | Predictable song composition |
| Cover feature | Extracts melody and reinterprets style | Enables creative remixes |
| Prompt-based control | Describes mood and progression | Fine creative direction |
| Low-end optimization | Improved bass and drums | Better listening experience |
| Auto lyrics | Generates lyrics from prompt | Faster ideation |
Creators can generate original background music tailored to specific scenes or moods, avoiding licensing constraints and repetitive stock tracks.
Developers can produce dynamic soundtracks that match gameplay intensity, from ambient exploration to high-energy combat sequences.
Applications can generate playlists or songs tailored to user preferences, mood, or context, rather than relying on pre-existing catalogs.
Artists and hobbyists can prototype musical ideas quickly, explore genre blending, or reinterpret existing melodies in new styles.
Music 2.6 is designed to work well within agent-based systems. It can be combined with higher-level logic that interprets user intent and generates music automatically based on context or user behavior.
Developers can build pipelines where music is generated dynamically, for example, adapting soundtracks in real time based on user input or environmental signals.
Measured on the AI/ML API gateway across live traffic. Figures are medians unless stated. Last updated 2026-10-12.
Use it for creating full-length songs from text prompts and lyrics, including vocals, instrumentation, and professional mixing.
Use it for AI cover generation, where an existing song is transformed into a new style while preserving its original melody.
Use it for lyric generation, since the model can automatically write lyrics from a prompt through its Lyrics Optimizer.
Use it for instrumental-only background music, with no lyrics needed.
import requests, time headers = {"Authorization": "Bearer " + AIMLAPI_KEY} job = requests.post( "https://api.aimlapi.com/v2/generate/audio", headers=headers, json={ "model": "minimax/music-2.6", "prompt": "Upbeat lofi background music" }, ).json() gid = job["generation_id"] while True: res = requests.get(f"https://api.aimlapi.com/v2/generate/audio?generation_id={gid}", headers=headers).json() if res.get("status") in ("completed", "error", "failed"): break time.sleep(3) print(res)
const headers = { Authorization: `Bearer ${process.env.AIMLAPI_KEY}`, "Content-Type": "application/json", }; const job = await (await fetch("https://api.aimlapi.com/v2/generate/audio", { method: "POST", headers, body: JSON.stringify({ "model": "minimax/music-2.6", "prompt": "Upbeat lofi background music" }), })).json(); let res; do { await new Promise((r) => setTimeout(r, 3000)); res = await (await fetch(`https://api.aimlapi.com/v2/generate/audio?generation_id=${job.generation_id}`, { headers })).json(); } while (!["completed", "error", "failed"].includes(res.status)); console.log(res);
# submit the job — the response contains "generation_id" curl -X POST https://api.aimlapi.com/v2/generate/audio \ -H "Authorization: Bearer $AIMLAPI_KEY" \ -H "Content-Type: application/json" \ -d '{"model":"minimax/music-2.6","prompt":"Upbeat lofi background music"}' # then poll for the result until it is ready curl "https://api.aimlapi.com/v2/generate/audio?generation_id={generation_id}" -H "Authorization: Bearer $AIMLAPI_KEY"
OpenAI-compatible — swap the base URL and it works with your existing SDK.
| Type | Price |
|---|---|
| Output | |
| Model | Input | Output | Context | Best for |
|---|---|---|---|---|
Music 2.6 This page | Audio generation | |||
| Speech synthesis | ||||
| Speech synthesis | ||||
| Speech synthesis | ||||
| Speech synthesis |
Music 2.6 takes audio, text as input and returns audio.
Music 2.6 became available on April 13, 2026.
Music 2.6 is priced at $0.195 / gen.
Music 2.6 is billed per generation — a fixed charge per output rather than by prompt length.
Music 2.6 was built by MiniMax.
Use minimax/music-2.6 as the model id on AI/ML API.
Yes. Music 2.6 is served through AI/ML API, so the same key and endpoint format used for other models applies.
