Quiver
Foundation models that generate and vectorise editable SVG
Quiver builds vector-native generative models (the Arrow family: 1.0, 1.1 and 1.1 Max) that turn a text prompt or a raster image into a production-ready, editable SVG rather than a traced approximation. Shipping today are logo generation and editable vector illustration, plus an API for listing models, text-to-SVG and image-to-SVG; typography and SVG animation are advertised as coming soon. It runs on a weekly credit pool with a free tier, has partnerships with Paper, Rive and Linearity, and is a16z-backed on an $8.3M seed. For brand and product designers who want a first vector pass they can open in a real editor.
Use the image-to-SVG endpoint when a user hands you a PNG logo or spot illustration and the deliverable must be a clean, editable path set; that is where it beats a general image model, which returns raster you then have to trace. Do not promise animated SVG or custom letterforms from it yet, both are still marked coming soon, and remember API credits are billed separately from the app's weekly pool, so a scripted batch will not draw on the free tier. Expect beta-grade output and keep a human in the loop before anything ships as a brand mark.
Why it is here. Vector-native generation is a real gap in the AI image stack, and Quiver is the best-funded attempt to solve it at the model layer instead of by tracing bitmaps.
- Arrow vector foundation models
- text-to-SVG and image-to-SVG API
- editable paths, not traced bitmaps
- $8.3M a16z seed
- typography and animation still unreleased
Verified 2026-08-15. HTTP 200. https://quiver.ai/ , https://quiver.ai/pricing , https://docs.quiver.ai/ (/developers 404s)
