Superwhisper, a voice AI startup, has launched its new S1 family of models on August 19, 2026, aiming to advance speech-to-text (STT) technology through both cloud and on-device solutions. Featuring models like the S1-Voice, a cloud-based STT model, and S1-mini, an on-device processing model, Superwhisper is pushing the boundaries of transcription speed and flexibility. Key to this launch is the partnership with Canadian AI developer Cohere, strengthening Superwhisper's capabilities in offline dictation and real-time processing.

The S1-Voice model stands out for its remarkable transcription speed, achieving capabilities up to 46 times faster than the original speech duration. This model significantly reduces the delay in transcription, maintaining a minimal lag of just 0.32 seconds for dictations under 30 seconds. Through support for over 100 languages, it provides robust multilingual transcription options — a testimony to Superwhisper's focus on inclusivity in speech recognition. This broad language support is further enhanced by Cohere Transcribe, which offers offline dictation services in 14 languages, as Slator reports.

The S1-mini model emerges as a specialized tool designed for on-device processing, with a parameter count of 0.6 billion. Its functionality is streamlined, focusing on a single task without following general instructions, allowing users to control it through a simple input line. This approach contrasts with typical chat models, emphasizing targeted efficiency over broader conversational capabilities. As Superwhisper highlights, "S1-mini is not a chat model and will not follow general instructions; it does one job, and you steer it with a control line at the top of the input." This specialization offers a significant advantage for users needing quick and efficient local processing without the need for cloud connectivity.

Superwhisper's strategic partnership with Cohere, initiated in July 2026, aligns with its goal to enhance language support and processing efficiency. The collaboration leverages Cohere's strengths in AI development, particularly in supporting offline capabilities that are crucial for areas with limited internet access. With the S1 family of models' debut, Superwhisper not only expands its technological reach but also sets a new standard in both speed and versatility for STT applications. The integrated approach of cloud-based speed and on-device reliability offers a glimpse into the future of voice AI technologies, where efficient multilingual support and user-centric design are paramount.