MIREX 2026 Music Evaluation via CMI-RewardBench
Models choose which of two generated recordings listeners prefer for a prompt. Organizers rank systems by accuracy on 249 pairs and report vocal and instrumental results separately.
MIREX ran the shared task. Its results include CloudEar and MuseCritic, which uses the MOSS-Audio-8B-Instruct audio language model.
- Kind
- Audio understanding and voice
- Level
- Everyday
- Scored
- Checked automatically
- Answers
- Multiple choice
- Tasks
- Public, plus a private test set
- Measures
- Capability
- Subject
- Arts
- Language
- English
- Interaction
- Single turn
- Items from
- Another benchmark, Generated, Written by crowdworkers
- Refresh
- One fixed set
- Run by
- A host
- Based on
- CMI-RewardBench, Music Arena
- In views
- Audio and speech
CloudEar-overall tops the leaderboard with 73.9%
Best score per entry from the benchmark's own leaderboard, read on November 20th, 2026. Where a model ran with several settings or harnesses, we keep its best. Names in gray aren't in our model list: fine-tunes, agent systems, products and older models. They don't count toward leaders.
- 73.9 CloudEar-overall
- 72.7 SongEval (Average)
- 72.3 first_submission
- 69.9 CloudEar (Musicality)
- 69.5 CloudEar (Alignment)
- 69.1 SongBench (Average)
- 69.1 SongBench (Musicality)
- 65.9 Baseline: CMI-RM
- 65.5 aimusic
- 64.3 TuneJury
- 60.2 Flip of "SYSTASYS Resource-Lite V2 - Blind Freeze 2026"
- 59.0 Random guess (always B)
- 55.4 MuseCritic
- 39.8 SYSTASYS Resource-Lite V2 - Blind Freeze 2026
- 1 CloudEar-overall 73.9%
- 2 SongEval (Average) 72.7%
- 3 first_submission 72.3%
- 4 CloudEar (Musicality) 69.9%
- 5 CloudEar (Alignment) 69.5%
- 6 SongBench (Average) 69.1%
- 7 SongBench (Musicality) 69.1%
- 8 Baseline: CMI-RM 65.9%
- 9 aimusic 65.5%
- 10 TuneJury 64.3%
- 11 Flip of "SYSTASYS Resource-Lite V2 - Blind Freeze 2026" 60.2%
- 12 Random guess (always B) 59%
- 13 MuseCritic 55.4%
- 14 SYSTASYS Resource-Lite V2 - Blind Freeze 2026 39.8%
Source: Official MIREX 2026 Music Evaluation via CMI-RewardBench leaderboard
| # | Entry | Score | Setting | Submitted by | Date | Link | Source |
|---|---|---|---|---|---|---|---|
| 1 | CloudEar-overall | 73.9% | User ID: 32 | futuremirex.com | Official leaderboard | ||
| 2 | SongEval (Average) | 72.7% | arxiv.org | Official leaderboard | |||
| 3 | first_submission | 72.3% | User ID: 59 | futuremirex.com | Official leaderboard | ||
| 4 | CloudEar (Musicality) | 69.9% | User ID: 32 | futuremirex.com | Official leaderboard | ||
| 5 | CloudEar (Alignment) | 69.5% | User ID: 32 | futuremirex.com | Official leaderboard | ||
| 6 | SongBench (Average) | 69.1% | arxiv.org | Official leaderboard | |||
| 7 | SongBench (Musicality) | 69.1% | arxiv.org | Official leaderboard | |||
| 8 | Baseline: CMI-RM | 65.9% | proceedings.mlr.press | Official leaderboard | |||
| 9 | aimusic | 65.5% | User ID: 68 | futuremirex.com | Official leaderboard | ||
| 10 | TuneJury | 64.3% | arxiv.org | Official leaderboard | |||
| 11 | Flip of "SYSTASYS Resource-Lite V2 - Blind Freeze 2026" | 60.2% | Official leaderboard | ||||
| 12 | Random guess (always B) | 59% | Official leaderboard | ||||
| 13 | MuseCritic | 55.4% | arxiv.org | Official leaderboard | |||
| 14 | SYSTASYS Resource-Lite V2 - Blind Freeze 2026 | 39.8% | User ID: 72 | futuremirex.com | Official leaderboard |
No lab model card reports it yet
We read the headline results of 540 launch posts from 57 labs, and none of them list this benchmark.
More in Audio and video
- Video-MMMU In use 300 expert-level lecture videos with questions testing perception, comprehension and adaptation of knowledge. Accuracy.
Qwen 3.8 Max 88.7%
- LVBench In use Hours-long video understanding with multiple-choice questions on reasoning, grounding and summarization.
Gemini 4 Argon 91.7%
- MMVU In use 3,000 expert-annotated questions on videos across 27 disciplines. Multiple-choice and open-ended accuracy.
Qwen 3.8 Max 82.4%
- MLVU In use Multi-task long video understanding (minutes to hours) with multiple-choice and generation tasks. Accuracy.
Qwen 3.8 Max 90.8%
- FLEURS In use Models transcribe read speech in 102 languages built from FLoRes-101 sentences. Model cards usually report word error rate, so lower scores are better.
Phi-4-multimodal 4.00%
- VoiceBench In use Spoken instructions with speaker, environment and content variation. Voice assistant responses scored by GPT judge or rules.
Qwen3.5-Omni-Plus 93.1%
Data audit
All 8 checks pass. Every benchmark
Catalog record 2 pass
- Pass Task access Public tasks, plus a private test set.
- Pass Confidence Checked against a primary source.
Scores 2 pass 3 notes
- Pass Independent runs Scores from the official leaderboard (14 entries).
- Pass Models scored 14 entries on its leaderboard, none a model we track.
- Note Newest score No scores yet.
- Note Frontier models No models scored yet.
- Note Headroom No scores yet.
Leaderboard import 3 pass
- Pass Importer We read the board on futuremirex.com: 14 entries at the last read, none a model we track.
- Pass Last read November 20th, 2026.
- Pass Read worked The last read worked.
Consistency 1 pass 1 note
- Note Lab claims No model has both a lab number and an independent run.
- Pass Known issues No published audit found problems.