Mureka AI: Song Generator, Pricing, V9.5 and Suno Comparison
Mureka V9.5 ranks 3rd on the Artificial Analysis leaderboard right behind Suno. We review pricing from $10, commercial rights, features, and API access.
👁 118 reads
Mureka is an AI song generator developed by Chinese conglomerate Kunlun Tech and operated through its Singapore subsidiary Skywork AI. Its latest V9.5 model ranks 3rd on the Artificial Analysis vocal leaderboard, trailing right behind two Suno models. The Pro plan costs $10 per month ($8 per month when billed annually), offering up to 500 songs along with commercial licensing rights, while the Free plan allows 4 songs daily on an older model without download capabilities.
Status as of October 8, 2026
- Model: V9.5 (available on the web service since mid-August 2026, via API since August 28, 2026). Also selectable: V9, V8, O2, and V7.6.
- Leaderboard: 3rd place on AA-Music-Vocal v1.1 (1,103 points), 4th in the instrumental ranking (1,080 points).
- Web pricing: Free $0, Pro $10/mo, Premier $30/mo. When billed annually: $8 and $24/mo.
- Rights: On paid plans, you own the generated track; on the free tier, Mureka retains ownership (non-commercial use only with required attribution).
- Language coverage: The official supported singing list covers 10 primary languages.
- Apps: Web browser, iOS, and Android (App Store and Google Play).
- API: platform.mureka.ai starting at $0.03 per song; V9.5 at $0.15. Also accessible via fal.ai.
What is Mureka
Mureka launched on August 14, 2024, in Singapore alongside Melodio, an AI-generated music streaming service. The project is backed by Kunlun Tech (Kunlun Wanwei), a Shenzhen-listed group whose portfolio includes the Opera browser. The apps are published by SKYWORK AI PTE LTD, the group’s AI division.
It operates much like Suno: you input lyrics or a descriptive prompt, select a style, and receive a fully produced song with vocals. However, Mureka bundles several features that competitors either lack or bury deep in complex menus:
- Reference track - upload an audio file and the model captures its genre style, vibe, and energy profile.
- Melody input - hum or record a short musical motif, and Mureka arranges instrumentation around it.
- Custom voices - select from preset vocalists or clone a voice from an audio sample (professional voice cloning is unlocked in the Premier plan).
- Stem splitter - separate generated audio into up to 12 individual stems, with MIDI and DAW session export (Premier).
- Remix, Extend, and inpainting - change genres, extend tracks, or regenerate a specific line of lyrics or melody snippet.
- Music videos - generate visualizers in one click or configure storyboard-guided clips alongside synced lyric videos.
- Speech - text-to-speech generation (the Pro plan allows up to 250 minutes of speech per month in place of music generations).
According to the official pricing page, over 200,000 users have subscribed to paid Pro and Premier tiers.
Mureka on the Artificial Analysis leaderboard
Artificial Analysis runs an independent platform where human evaluators compare randomized, blind pairs of music generations across models. Version v1.1 of their music leaderboard (as of roughly October 7, 2026) is based on 1,000 standardized prompts spanning 17 genres and approximately 37,000 blind panel votes. Results use an Elo rating scale anchored by MiniMax Music 3.0 at 1,000 points. A wider breakdown is available in our AI music generator rankings.
| Model | Vocal (Elo) | Rank | Instrumental (Elo) | Rank |
|---|---|---|---|---|
| Suno v6 | 1142 | 1 | 1140 | 1 |
| Suno v6-mini | 1116 | 2 | 1109 | 2 |
| Mureka V9.5 | 1103 | 3 | 1080 | 4 |
| Mureka V9 | 1052 | 4 | 1081 | 3 |
| Lyria 3.5 | 1052 | - | 1076 | - |
| Lyria 3 Pro | 1046 | - | 1078 | - |
| Eleven Music v2.5 | 1012 | - | - | - |
| MiniMax Music 3.0 | 1000 (anchor) | - | - | - |
- The 95% confidence interval for V9.5 in vocal generation spans 1,088-1,118 (across 2,112 samples), meaning its statistical lead over Suno v6-mini is not absolute.
- In instrumental music, V9.5 does not outperform V9 (1,080 vs 1,081), while Lyria sits just 2 to 4 points behind.
Mureka also highlights a 1st place finish in the “A Thousand Listeners” benchmark hosted on aimusicbenchmark.org (snapshot from September 1, 2026). In that test, 1,000 human listeners cast 7,124 ratings, placing Mureka V9.5 at a 65.6% win rate ahead of Suno at 62.5%. A few caveats apply:
- The benchmark tested older Suno V5.5 models, not Suno v6,
- In direct head-to-head match-ups, Mureka edged past Suno in only 52.4% of matchups,
- The testing body has not disclosed who funded the study or the sourcing methodology for the listener panel.
How to get started with Mureka step by step
- Visit mureka.ai or install the Mureka app (published by Skywork AI) from the App Store or Google Play, then register an account.
- Click Create. In simple mode, type a descriptive concept for your track; in custom mode, paste your written lyrics and add structural markup tags like
[Verse],[Chorus], and[Bridge]. - Specify stylistic tags: genre, tempo, mood, singer gender, and vocal timbre (e.g. “indie pop, 110 BPM, female vocal, warm, airy”).
- Select your model. Free accounts are limited to v7.5-all; V9.5 requires an active Pro subscription.
- Optionally attach an audio reference track (Reference), record a melodic humming guide, or select a specific voice profile.
- Generate your options and listen through the variations. If one line misses the mark, highlight that passage to inpaint and regenerate that section rather than re-rolling the whole song.
- Download your finished audio (MP3 on Pro; lossless WAV, individual stems, and MIDI on Premier) or generate an accompanying video visualizer.
Mureka model version history
Release dates below correspond to the official API changelog. Web interface availability often preceded API launches, as noted.
| Model | Release date | Key additions |
|---|---|---|
| SkyMusic 2.0 | August 14, 2024 | Initial launch of Mureka and Melodio |
| V6 and O1 | Late March 2025 (web), API: V6 June 3, 2025; O1 July 29, 2025 | O1 marketed as the first “reasoning” music model |
| V7 / V7.5 | July 29, 2025 / September 25, 2025 | Audio fidelity upgrades; V7.5 targeted at Chinese vocals; v7.5 powers the current Free plan |
| V7.6 and O2 | December 9, 2025 | WAV export via API; dual voice and prompt guidance across 10 languages added January 5, 2026 |
| V8 | API March 2, 2026 (web press reports suggested January 28, 2026 - unverified) | Production-ready mixing aesthetic |
| V9 | Late March 2026 (web), API April 9, 2026 | Paragraph-level lyrical steering, cleaner mix dynamics, accelerated generation times |
| V9.5 | Unveiled at WAIC (July 2026), web around August 13, 2026, API August 28, 2026 | More natural vocal articulation, stricter adherence to genre parameters and descriptions |
Alongside V9.5, Mureka announced an upcoming O3 model engineered to “deliberate” musical structure during the generation process. As of October 8, O3 is absent from the active model selector; the platform lists V9.5, V9, V8, O2, and V7.6. The developer claimed on August 31, 2026, that V9.5 achieves 97% prompt adherence - note that this figure comes from internal evaluations rather than third-party audits.
How much does Mureka cost
| Plan | Song allowance | Available models | Downloads and commercial rights | Pricing (mureka.ai) |
|---|---|---|---|---|
| Free | 4 daily, 2 concurrent in shared queue | v7.5-all | No downloads, non-commercial only | $0 |
| Pro | 5,000 Gold = up to 500 songs/mo, 10 concurrent | V9.5, V9, V8, O2, V7.6 | Unlimited MP3, commercial rights, video visualizers, Reference, Remix, Extend | $10/mo or $96/yr ($8/mo) |
| Premier | 20,000 Gold = up to 2,000 songs/mo, 10 concurrent | Same as Pro | Pro features plus WAV, 12 stems, MIDI, DAW sessions, Mureka Studio, voice cloning | $30/mo or $288/yr ($24/mo) |
Pricing verified directly on mureka.ai/subscribe on October 8, 2026. Gold credit top-ups are restricted to paying subscribers. Mobile app subscription prices can vary based on app store tiers; verify current figures during checkout on your platform.
API access via platform.mureka.ai uses prepaid credits valid for 12 months. An initial starter package costs $10 (limited to one purchase per account), with subsequent credit tiers starting at $1,000. Generating a song from supplied lyrics costs $0.03 (V7.6), $0.045 (V8, V9, O2), or $0.15 (V9.5). Generating from a purely descriptive prompt runs $0.30, or $0.50 on V9.5. Custom voice cloning costs $5 per voice model, and 12-stem isolation costs $0.70 per song. On fal.ai, Mureka 9.5 is billed at $0.225 per song with lyrics and $0.75 for prompt-only song generation.
Copyright and commercial usage
Mureka’s terms of service, dated May 13, 2026, delineate rights based on account tier:
- Paid subscriptions and purchased API credits - you own the intellectual property rights in the generated sound recordings. You are permitted to distribute them on commercial streaming services, license them for advertising, or monetize them in video productions. Upgrading to a paid plan retroactively grants commercial rights over tracks generated earlier on the account.
- Free tier - Mureka retains all ownership rights. Users receive a limited license restricted to non-commercial internal evaluation, provided Mureka is explicitly credited.
Two essential legal provisions apply to all tiers. Mureka reserves a royalty-free, perpetual license to host and utilize your outputs, including for commercial and training purposes. The provider also does not guarantee novelty, non-infringement, or that the resulting compositions can be successfully registered with local copyright offices. We explore competitor licensing frameworks in our guide to Suno copyright rules.
Language support and international singing
The web interface supports 11 display languages. The official list of explicitly supported singing languages includes 10 entries: Chinese, English, Japanese, Korean, Portuguese, Spanish, German, French, Italian, and Russian. Other languages are not officially validated, and API documentation notes that unlisted languages “may lead to unexpected generation errors.”
While unlisted languages can be pasted into the lyric box and the model will attempt phonetic vocalization, doing so falls outside official performance specifications.
Our benchmark notes: why this round skipped manual audio runs
During our October 8, 2026 music generator benchmark, we tested every accessible model with identical song lyrics (two verses and a chorus totaling 52 words). Mureka was omitted from that specific comparative audio batch for practical reasons:
- The official API (platform.mureka.ai) requires an account setup and an upfront $10 starter credit block.
- fal.ai hosts Mureka 9.5 at $0.225 per lyrical song, but our fal.ai balance was depleted on the test date.
- The web Free plan is restricted to model v7.5-all and disallows audio downloads, preventing consistent transcription measurement using our standard Demucs stem isolation and Whisper large-v3 automated pipeline.
For context: in that identical comparative test run, Eleven Music v2.5 and Lyria 3 Pro accurately enunciated 103 out of 104 tested words, while the locally hosted open-weight ACE-Step 1.5 hit 101. Full data tables are tracked in our AI music generator rankings. We will integrate Mureka V9.5 audio samples in an upcoming benchmark revision.
Mureka vs the competition
| Feature | Mureka | Suno | Udio | Eleven Music | Lyria |
|---|---|---|---|---|---|
| Latest model | V9.5 (08.2026) | v6 (09.2026) | v1.5 Allegro (03.2025) | Music v2.5 (09.2026) | Lyria 3.5 (07.2026) |
| AA-Music-Vocal v1.1 | 1,103 (3rd place) | 1,142 (1st place) | Not ranked in checked data | 1,012 | 1,052 |
| Free plan | 4 songs/day, no downloads | Yes | 10 credits/day, no downloads | Yes | Not verified |
| Paid tier starting price | $10/mo | Varies by app store / web | Varies by platform | $6/mo | $6/mo (Flow Music) |
| Commercial rights | Yes, Pro tier and above | Yes, paid plans | No | Yes, Starter and above | Unconfirmed |
| API access | Yes, native API & fal.ai | No | No | Yes | Via Gemini API |
Mureka’s primary advantage is its volume-to-price ratio: 500 songs for $10 with audio quality directly competitive with Suno. Udio disabled downloads in late 2025 (see our analysis of Udio vs Suno). For a wider assessment covering open models like Stable Audio, consult our guide to AI music generators.
Who is Mureka best for?
- Volume creators: If you produce background cues for video production, retail sound design, advertising, or podcast stingers, 500 songs for $10 represents the most cost-effective entry point among top-tier leaderboard models.
- DAW producers: If you need separate multi-track stems, MIDI transcriptions, and raw arrangements for production inside Ableton or Logic Pro (Premier plan).
- Users needing unsupported singing languages: If your project demands flawless pronunciation in niche or non-standard languages outside the 10 supported tongues, dedicated alternatives like Eleven Music may prove more dependable.
- Zero-budget distribution: Free-tier outputs cannot be downloaded or published commercially since Mureka retains intellectual property ownership.
Frequently asked questions
What is Mureka AI?
Mureka is an AI music production platform created by Skywork AI (Kunlun Tech). It synthesizes complete vocal songs, instrumental tracks, video visualizers, and spoken audio from text lyrics or descriptive prompts across web, iOS, and Android.
Can you use Mureka for free?
Yes, a Free plan offers 4 songs per day using the v7.5-all legacy model. Audio downloads are disabled on the free tier, and full intellectual property rights remain with Mureka. Accessing the flagship V9.5 model, file exports, and commercial ownership requires the Pro plan.
How much does Mureka cost?
On mureka.ai, the Pro plan is priced at $10 per month (or $8 per month billed annually at $96). The Premier plan costs $30 monthly ($24 monthly on annual billing at $288). Mobile app pricing may differ based on local app store pricing structures.
Is Mureka better than Suno?
Suno v6 holds higher objective marks on the Artificial Analysis leaderboard (1,142 vs 1,103 in vocal Elo). However, Mureka provides vastly higher generation quotas per dollar, offers official developer APIs, and exports up to 12 separate stems with MIDI. For raw output quality, choose Suno; for volume and DAW workflow integration, Mureka offers stronger utility.
What languages does Mureka support?
Mureka officially supports singing in 10 languages: Chinese, English, Japanese, Korean, Portuguese, Spanish, German, French, Italian, and Russian. While other languages can be entered into lyric fields, pronunciation accuracy is not officially guaranteed.
Can you monetize music made with Mureka?
Yes. Subscribing to paid plans or using purchased API credits assigns full commercial ownership to you, including retroactive rights for tracks generated previously on the same account. Free-tier generations are strictly licensed for non-commercial personal evaluation and require creator attribution.
Google · Your sourcesQuantslant higher in your Google - one clickAdd to preferred sources →Sources and verification dates
- Mureka: pricing and subscription FAQ - verified October 8, 2026.
- Mureka API: API pricing, model changelog, documentation FAQ - verified October 8, 2026.
- Mureka Terms of Service dated May 13, 2026, section on intellectual property rights.
- Artificial Analysis: Vocal Leaderboard and Instrumental Leaderboard v1.1 - verified October 8, 2026.
- aimusicbenchmark.org, “A Thousand Listeners” (snapshot September 1, 2026) - verified October 8, 2026.
- Mureka V9.5 Press Announcement (August 31, 2026), Variety coverage on V9.5 and O3.
- Kunlun Tech launch announcement for Mureka and Melodio (August 14, 2024), MBW on O1 reasoning model (March 2025).
- fal.ai: Mureka 9.5 model endpoint - verified October 8, 2026.
- App Store (application “Mureka: AI Radio & Music Maker”, SKYWORK AI PTE LTD) and Google Play (Skywork AI Pte.Ltd.) listings - verified October 8, 2026.
More on this from QUANTSLANT
ai tools·Perplexity Pro Search with Real-Time Data: A Real Google Threat?Perplexity Pro adds real-time data streaming under 2 seconds. Can live stock prices, weather, and social trends let it challenge Google and ChatGPT?

