Original example and model/program roles
After transcription: connect a real device →
Whistle adds speech input: its 16.9 MB model transcribes audio on the CPU, then Needle selects a call using tool definitions.
Cactus Compute · Released 2026-10-02
Cactus Whistle
A 16.9 MB CPU speech model with seven supported languages, excluding Chinese. It can share an engine with Needle 3 to return text and tool arguments. Speech error rates have a separate scope from our text, coding and physics boards.
Original model explanation ↗No comparable leaderboard score is included yet. Explore the evaluations and examples below.
Evaluations: conditions and results
Vendor tests; some baseline error rates quoted from their authors
Speech error rates and M4 Pro CPU tests ↗
Official method, not a score in our three boards
Languages, keywords and arguments after transcription ↗
How this capability connects to works and uses
Official JSON example; a real lighting interface needs separate integration
How an audio clip returns tool calls ↗
Cactus Compute model evolution
- Whistle · 原发布
A 16.9 MB speech model supports seven languages and shares a CPU engine with Needle 3; tool-call JSON still needs software connected to services or devices.
Original release ↗