desertant.com

mx

smtp.google.com

ns

quincy.ns.cloudflare.com

tara.ns.cloudflare.com

caa

txt

google-site-verification=HKpI6FZkc_YFHIRia9lnnASRBpom1dPIAPZrDbRsMtk

ab2zcm0ly5

google-site-verification=rEoF6dTD5jPaHS2vrujgVGj6IsOvvcqZN29eiQ86ST0

v=spf1 include:_spf.google.com ~all

dmarc

v=DMARC1; p=none;

# robots.txt for desertant.com
# Static marketing site. Everything is public and welcome to crawlers.
# LLM index: https://desertant.com/llms.txt

User-agent: *
Allow: /

# Content usage preferences (contentsignals.org).
# search   = appear in search indexes and results
# ai-input = may be used as input for AI answers (RAG, assistants)
# ai-train = may be used to train AI models
# All three on: this is public marketing content and we want it discoverable.
# The models and weights stay under their own license on Hugging Face / GitHub.
Content-Signal: search=yes, ai-input=yes, ai-train=yes

Sitemap: https://desertant.com/sitemap.xml

# Desert Ant Labs

> The fastest models for your task, running on-device. Each model is free up to 100k monthly active devices. Inference is unlimited.

## Models (available)

- [Redact: PII redaction](https://desertant.com/models/redact/): On-device redaction that strips personal data as you type, in 27 languages, so it never reaches your servers or logs. Platforms: Swift, Kotlin, JS. Sources: [Hugging Face](https://huggingface.co/desert-ant-labs/redact), [SDK](https://github.com/Desert-Ant-Labs/desert-ant-core), [Docs](https://desertant.com/docs/redact/).
- [Emo: Emoji suggestions](https://desertant.com/models/emo/): Type a task or a message and get the emoji that fit, in 22 languages, on the device. Platforms: Swift, Kotlin, JS. Sources: [Hugging Face](https://huggingface.co/desert-ant-labs/emo), [SDK](https://github.com/Desert-Ant-Labs/desert-ant-core), [Docs](https://desertant.com/docs/emo/).
- [Gist: Content topic tagging](https://desertant.com/models/gist/): On-device topic tagging that reads a title or a post and tells you its topics, in 101 languages, with no server call. Platforms: Swift, Kotlin, JS. Sources: [Hugging Face](https://huggingface.co/desert-ant-labs/gist), [SDK](https://github.com/Desert-Ant-Labs/desert-ant-core), [Docs](https://desertant.com/docs/gist/).
- [Shapes: Shape recognition](https://desertant.com/models/shapes/): On-device shape recognition that turns one hand-drawn stroke into a rectangle, ellipse, triangle, star or line in under 10ms. Platforms: Swift, Kotlin, JS. Sources: [Hugging Face](https://huggingface.co/desert-ant-labs/shapes), [SDK](https://github.com/Desert-Ant-Labs/desert-ant-core), [Docs](https://desertant.com/docs/shapes/).
- [Clear: Speech enhancement](https://desertant.com/models/clear/): On-device speech enhancement that gives your recording the warm, close-miked sound of a podcast studio, processed 100% on-device. Platforms: Swift, Kotlin, JS. Sources: [Hugging Face](https://huggingface.co/desert-ant-labs/clear), [SDK](https://github.com/Desert-Ant-Labs/desert-ant-core), [Docs](https://desertant.com/docs/clear/).
- [Uhm: Filler-word detection](https://desertant.com/models/uhm/): On-device filler detection that marks every um, uh and hmm to within 20ms, an hour of audio in 12s. Platforms: Swift. Sources: [Hugging Face](https://huggingface.co/desert-ant-labs/uhm), [SDK](https://github.com/Desert-Ant-Labs/desert-ant-core), [Docs](https://desertant.com/docs/uhm/).
- [Tongue: Language identification](https://desertant.com/models/tongue/): On-device language detection for text across 84 languages. Identify the language in tens of microseconds, from just a few words. Platforms: Swift, Kotlin, JS. Sources: [Hugging Face](https://huggingface.co/desert-ant-labs/tongue), [SDK](https://github.com/Desert-Ant-Labs/desert-ant-core), [Docs](https://desertant.com/docs/tongue/).
- [Align: Word timestamps](https://desertant.com/models/align/): Word timestamps on device, for any transcriber. Cut, caption, and highlight on the word, five times more accurately, from 0.7MB. Platforms: Swift, JS. Sources: [Hugging Face](https://huggingface.co/desert-ant-labs/align), [SDK](https://github.com/Desert-Ant-Labs/desert-ant-core), [Docs](https://desertant.com/docs/align/).
- [Clips: Clip selection](https://desertant.com/models/clips/): Clip selection that turns a long recording into ranked shorts and highlights from a transcript. On-device on iPhone or Mac. Platforms: Swift. Sources: [Hugging Face](https://huggingface.co/desert-ant-labs/clips), [SDK](https://github.com/Desert-Ant-Labs/desert-ant-core), [Docs](https://desertant.com/docs/clips/).
- [Ear: Spoken language detection](https://desertant.com/models/ear/): On-device spoken language identification across 99 languages, from a short stretch of audio, before transcription starts. Platforms: Swift, Kotlin, JS. Sources: [Hugging Face](https://huggingface.co/desert-ant-labs/ear), [SDK](https://github.com/Desert-Ant-Labs/desert-ant-core), [Docs](https://desertant.com/docs/ear/).
- [Title: Titles and descriptions](https://desertant.com/models/title/): On-device titles and descriptions: a short factual title and a one or two sentence description for any passage, on Apple silicon. Platforms: Swift. Sources: [Hugging Face](https://huggingface.co/desert-ant-labs/title), [SDK](https://github.com/Desert-Ant-Labs/desert-ant-core), [Docs](https://desertant.com/docs/title/).
- [Voz: Speech recognition](https://desertant.com/models/voz/): Transcribe 10 minutes of audio in 2s on an iPhone, 14x faster than Whisper, accurate word timestamps. Platforms: Swift, JS. Sources: [Hugging Face](https://huggingface.co/desert-ant-labs/voz), [SDK](https://github.com/Desert-Ant-Labs/desert-ant-core), [Docs](https://desertant.com/docs/voz/).

## Models (closed beta)

- [Moderator: Content moderation](https://desertant.com/models/moderator/): On-device NSFW detection that flags nudity before upload, trained only on licensed and synthetic data, safe to ship commercially. Sources: [Hugging Face](https://huggingface.co/desert-ant-labs/moderator), [SDK](https://github.com/Desert-Ant-Labs/desert-ant-core), [Docs](https://desertant.com/docs/moderator/).
- [Schemer: Structured extraction](https://desertant.com/models/schemer/): Extract typed fields from any text against your own schema, on the device, with a field left empty where the text says nothing. Sources: [Hugging Face](https://huggingface.co/desert-ant-labs/schemer).
- [Eye: Image and video scoring](https://desertant.com/models/eye/): On-device frame scoring that tells you which shot to keep from a burst or a clip, what is in the shot, and which shots repeat.
- [Who: Who said what](https://desertant.com/models/who/): On-device speaker labeling that turns a recording into per-person turns with face boxes and timestamps, ready to cut or caption.
- [Face: Face matching](https://desertant.com/models/face/): On-device face matching that finds the same person across your photo library or through a video, with no image leaving the device.
- [Toxic: Hate speech triage](https://desertant.com/models/toxic/): On-device moderation that flags hateful, abusive and threatening text in 23 European languages, before the message is sent. Sources: [Hugging Face](https://huggingface.co/desert-ant-labs/toxic).

## Platforms

- [Swift (iOS, macOS)](https://desertant.com/swift/): Drop-in Swift packages that run small, specialized models on device, on iOS and macOS, in a few lines of code on Core ML.
- [Kotlin (Android, JVM)](https://desertant.com/kotlin/): Kotlin libraries that run models on device across Android and the JVM, on LiteRT and ONNX Runtime.
- [JavaScript / TypeScript (Web, Node.js)](https://desertant.com/js/): npm packages that run models in the browser and in Node, with TypeScript types included.

## Resources

- [Machine-readable catalog](https://desertant.com/catalog.json): all models and platforms as JSON.
- [Full content](https://desertant.com/llms-full.txt): every model overview and use case in one file.
- [SDKs on GitHub](https://github.com/Desert-Ant-Labs)
- [Model weights on Hugging Face](https://huggingface.co/desert-ant-labs)