| Name | Modified | Size | Downloads / Week |
|---|---|---|---|
| Parent folder | |||
| README.md | 2026-11-12 | 1.3 kB | |
| v0.2.0 source code.tar.gz | 2026-11-12 | 91.2 kB | |
| v0.2.0 source code.zip | 2026-11-12 | 114.6 kB | |
| Totals: 3 Items | 207.1 kB | 0 | |
What's New
The Wiro AI provider now runs image, speech, transcription and video models as well as chat, through the AI SDK's own functions:
- Images:
wiro.image(modelId)forgenerateImage, with text to image and image editing fromprompt.imagesandprompt.mask. - Speech:
wiro.speech(modelId)forgenerateSpeech, with each model's voices and languages. - Transcription:
wiro.transcription(modelId)fortranscribe. A public audio URL goes straight to Wiro, and timings come back when the model writes them. - Video:
wiro.video(modelId)forexperimental_generateVideo,experimental_startVideoandexperimental_getVideoStatus, from text or a start image, with webhooks.
size, aspectRatio, seed, n, voice, language, outputFormat, resolution, duration and the other AI SDK options go to each model's own settings, whatever the model calls them. Model settings of your own go in providerOptions.wiro.
Every result names its Wiro task and what it cost in providerMetadata.wiro, and WiroTaskError tells you which task gave no result and why. Signature projects keep using API_KEY:API_SECRET. The README has an example for each capability, and examples/ has runnable files.
Install: npm i @wiro-ai/wiro-llm ai