Whistle
#9 in Transcription (STT)Unknown · since 2. Oktober 2026 · 2× · last seen Oct 11, 2026
40
Momentum
Whistle is a speech-to-text (STT) model from Cactus Compute, released on October 2, 2026 as a single 16.9 MB file that runs on CPU without dependencies. It transcribes up to 30 seconds of 16 kHz mono audio per pass in seven languages and returns word timestamps and speech embeddings. The benchmarks are vendor-reported and not independently verified.
Momentum trend
13.07.11.10.
Features
| Real-Time Streaming | Not supported (streaming output reportedly planned) |
| Latency | 11.1 ms time to first token (10 s audio, Apple M4 Pro, vendor-reported) |
| License | Apache-2.0 |
| Platform | 17 platforms incl. macOS, Linux (x86-64, ARM64, ARMv7, RISC-V, MIPS32), Windows x64/ARM, Android, iOS, watchOS, tvOS, browser (WebAssembly), WASI |
| Price | Free (per third-party source; no official pricing page found) |
| Release Date | October 2, 2026 |
| Languages | English, German, French, Spanish, Italian, Dutch, Polish (language auto-detected or user-specified) |