Local AI Registry

Leaderboard

QuALITY

quality
JSON APIBack to leaderboard search

Record

Category
knowledge
Scored models
33

Scores

ModelScore
Qwen/Qwen3.6-27B100
Qwen/Qwen3.6-35B-A3B100
Qwen/Qwen3.6-35B-A3B100
moonshotai/Kimi-K2.6100
nvidia/Nemotron-Labs-TwoTower-30B-A3B-Base-BF16100
Qwen/Qwen3.5-35B-A3B90.5
Qwen/Qwen3.6-27B89
nvidia/Cosmos-Predict2-2B-Sample-Action-Conditioned79.8
nvidia/Cosmos-Predict2-2B-Sample-Action-Conditioned70.4
nvidia/Cosmos-Predict2-14B-Video2World69.9
nvidia/Cosmos-Predict2-2B-Video2World69.6
nvidia/Cosmos-Predict2-2B-Sample-Action-Conditioned64.15
Qwen/Qwen-72B61.38
Qwen/Qwen-72B61.38
Qwen/Qwen-72B61.38
sarulab-speech/sidon-v0.148
nvidia/Harmonizer31.06
—24
—24
—24
—24
—24
—24
—24

Remaining fields

Identity, launch, related records, and measured speed are shown above. This is the rest of the normalized record.

category
knowledge
id
quality
name
QuALITY
schema version
local-ai-registry/v1

rows

variantconfcontextmodel idorgrankrootscore
Qwen3.6-27B-NVFP4mediumqualityqwen3-6-27bunsloth1Qwen/Qwen3.6-27B100
Qwen3.6-35B-A3B-NVFP4mediumqualityqwen3-6-35b-a3bunsloth2Qwen/Qwen3.6-35B-A3B100
Qwen3.6-35B-A3B-NVFP4-Fastmediumqualityqwen3-6-35b-a3bunsloth3Qwen/Qwen3.6-35B-A3B100
Kimi-K2.6-GGUFmediumqualitykimi-k2-6unsloth4moonshotai/Kimi-K2.6100
Nemotron-Labs-TwoTower-30B-A3B-Base-BF16mediumQuality retained (% of AR baseline)nemotron-labs-twotower-30b-a3b-base-bf16nvidia5nvidia/Nemotron-Labs-TwoTower-30B-A3B-Base-BF16100
Ising-Calibration-1-35B-A3BmediumQ4 Fit Quality Assessmentqwen3-5-35b-a3bnvidia6Qwen/Qwen3.5-35B-A3B90.5
Qwen-Image-BenchmediumQualityqwen3-6-27bQwen7Qwen/Qwen3.6-27B89
Cosmos-Predict2-2B-Sample-Action-ConditionedmediumQuality ScoreUnknownnvidia8nvidia/Cosmos-Predict2-2B-Sample-Action-Conditioned79.8
Cosmos-Predict2-2B-Sample-Action-ConditionedmediumImaging QualityUnknownnvidia9nvidia/Cosmos-Predict2-2B-Sample-Action-Conditioned70.4
Cosmos-Predict2-14B-Video2Worldhigh**PBench Quality Score**Unknownnvidia10nvidia/Cosmos-Predict2-14B-Video2World69.9
Cosmos-Predict2-2B-Video2Worldhigh**PBench Quality Score**Unknownnvidia11nvidia/Cosmos-Predict2-2B-Video2World69.6
Cosmos-Predict2-2B-Sample-Action-ConditionedmediumAesthetic QualityUnknownnvidia12nvidia/Cosmos-Predict2-2B-Sample-Action-Conditioned64.15
Qwen-72B-ChatlowQuALITYqwen-72bQwen13Qwen/Qwen-72B61.38
Qwen-72B-Chat-Int4lowQuALITYqwen-72bQwen14Qwen/Qwen-72B61.38
Qwen-72B-Chat-Int8lowQuALITYqwen-72bQwen15Qwen/Qwen-72B61.38
Sidon-CoreMLmediumqualityw2v-bert-2-0aufklarer16sarulab-speech/sidon-v0.148
HarmonizermediumHarmonizer: temporal mode<br>(highest output)<br><br>Inference enabled through the following checkpoint:<br>`diffusion_harmonizer.pkl`Unknownnvidia17nvidia/Harmonizer31.06
maya1mediumQualityUnknownmaya-research18—24
Orpheus-3b-FT-Q8_0.ggufmediumqualityUnknownlex-au19—24
Orpheus-3b-German-FT-Q8_0.ggufmediumqualityUnknownlex-au20—24
Orpheus-3b-French-FT-Q8_0.ggufmediumqualityUnknownlex-au21—24
Orpheus-3b-Hindi-FT-Q8_0.ggufmediumqualityUnknownlex-au22—24
Orpheus-3b-Korean-FT-Q8_0.ggufmediumqualityUnknownlex-au23—24
Orpheus-3b-Chinese-FT-Q8_0.ggufmediumqualityUnknownlex-au24—24
Orpheus-3b-FT-Q4_K_M.ggufmediumqualityUnknownlex-au25—24
maya1-GGUFmediumQualityUnknownMungert26—24
Orpheus-3b-Italian_Spanish-FT-Q8_0.ggufmediumqualityUnknownlex-au27—24
whisper-large-v3-high-mixed-nlmediumMid-High Quality Filtered + CVwhisper-large-v3yuriyvnv28openai/whisper-large-v35.7
whisper-large-v3-high-mixed-nlmedium**High-Quality Filtered + CV**whisper-large-v3yuriyvnv29openai/whisper-large-v35.2
mustangohigh**Audio Quality** ↑Unknowndeclare-lab30—3.35
vocalburst-captioning-whisperlowcaption qualitywhisper-smalllaion31laion/sound-effect-captioning-whisper3.24
Qwen3-TTS-12Hz-1.7B-Base-MLX-bf16mediumqualityUnknownaufklarer32Qwen/Qwen3-TTS-12Hz-1.7B-Base1.7
Ace-Step-1.5-ScragVAElowPhase 2 (Quality)Unknownscragnog33ACE-Step/ace-step-v1.5-1d-vae-stable-audio-format1.5
Provenance & metadata (1)

source

kind
leaderboard-scrape
paths
benchmarks/quality.html, docs/notes/hf-benchmark-contexts.json