Loading the catalogue…
Klang AI AB · Pianissimo family
pianissimo
Released 23 Sept 2026
Open-weight Swedish speech-recognition model from Klang AI that returns punctuated, capitalised text.
€0.0035per audio minute
Loading the catalogue…
Open-weight Swedish speech-recognition model from Klang AI that returns punctuated, capitalised text.
| Inference Provider | Sovereignty | Runs in | Price | Status | Pin |
|---|---|---|---|---|---|
| L1 | €0.0035 per audio minute | Online |
13 use cases classified by the curator. The note says why.
Agentic coding requires tool-use and multi-step execution, neither of which is documented for this draft variant.
No vision capability is flagged, so chart extraction from images is unsupported by the current profile. Revisit if Klang AI publishes multimodal support.
No vision capability is flagged, so the model has no documented path for OCR on scanned or image-based documents.
No function-calling support is flagged, so structured tool invocation cannot be assumed.
No vision capability is flagged on the draft pianissimo card, so it should not be routed to image question-answering. Klang AI has published no multimodal evidence.
The reported size of roughly 600M parameters is small by current standards, and no reasoning benchmarks have been published to support multi-step planning. Treat this as unverified.
Sourced numbers, not our own evaluations. Percentiles are against the transcription models on Stav.
No benchmark source yet
curl https://api.stav.ai/v1/audio/transcriptions \
-H "Authorization: Bearer $STAV_API_KEY" \
-F model="klang/pianissimo" \
-F file="@meeting.mp3"→ X-Stav-Request-Id · X-Stav-Model · X-Stav-Provider → X-Stav-Sovereignty-Level — the rung the request was served at