In March 2026, Cohere released Transcribe, a 2B Apache 2.0 model that took the top of the Hugging Face Open ASR Leaderboard at 5.42% average word error rate. Five weeks later, IBM shipped Granite Speech 4.1 2B at 5.33%. Since then, ARK-ASR-3B and MOSS-Transcribe-preview-2B have posted lower numbers, with the top of the leaderboard now separated by less than one WER point. However, the leaderboard average is not a single fixed quantity; Cohere’s 5.42% is an average across eight English test sets including TED-LIUM, while ARK-ASR-3B’s 5.04% is an average across seven sets with TED-LIUM absent. Recomputing Cohere’s scores over the same seven sets ARK reports lands Cohere at 5.84, and Granite Speech 4.1 2B moves from 5.33 to 5.65. The MOSS-Transcribe-preview-2B card states the model was fine-tuned with reinforcement learning on the Open ASR Leaderboard training splits. Additionally, when private held-back evaluation sets from Appen covering Australian, Canadian, Indian, and American accents are toggled on, zoom/scribe_v1 moves from #4 to #1.