Official benchmarks map / open-asr-leaderboard
Edge0/ARK-ASR-3B is ranked first on the open-asr-leaderboard leaderboard with 4.76, -2.3% ahead of OpenMOSS-Team/MOSS-Transcribe-preview-2B. Read live from the Hugging Face API on November 6, 2026.
| # | Model | Organization | Score |
|---|---|---|---|
| 1 | Edge0/ARK-ASR-3B | Edge0 | 4.76 |
| 2 | OpenMOSS-Team/MOSS-Transcribe-preview-2B | OpenMOSS-Team | 4.87 |
| 3 | OpenMOSS-Team/MOSS-Transcribe-Diarize | OpenMOSS-Team | 5.17 |
| 4 | hf-audio/MOSS-Transcribe-Diarize-HF | hf-audio | 5.17 |
| 5 | moondream/parakeet-ultra | moondream | 5.32 |
| 6 | CohereLabs/cohere-transcribe-03-2026 | CohereLabs | 5.42 |
| 7 | Edge0/ARK-ASR-0.6B | Edge0 | 5.53 |
| 8 | soundsgoodai/Zipformer-cr-ctc-transducer-XL-290M | soundsgoodai | 5.56 |
| 9 | Qwen/Qwen3-ASR-1.7B | Qwen | 5.76 |
| 10 | microsoft/Phi-4-multimodal-instruct | microsoft | 6.02 |
open-asr-leaderboard is a dataset that Hugging Face marks as an official benchmark. Its leaderboard is built automatically from the .eval_results files that model repositories publish, so entries are mostly self-reported by model authors and evaluation settings differ between submissions. A verified tag appears only where Hugging Face marks an entry as verified.
Original board: huggingface.co/datasets/hf-audio/open-asr-leaderboard
Part of the Hugging Face official benchmarks map by QuantID. Unofficial. Not affiliated with or endorsed by Hugging Face.