AI Models
Every AI feature on tomai.uk runs on a concrete, named model — no vague "AI technology" anywhere. This page lists each model, what it powers, and where it runs. We update it whenever the model behind a feature changes.
Last updated: 2026-08-15
Text AI: DeepSeek
All text-based AI tools — fortune readings (tarot, horoscopes, Zi Wei Dou Shu and more), the AI writing tools, the AI content detectors, and the prompt safety screening that guards AI image generation — are powered by DeepSeek’s deepseek-v4-flash model (DeepSeek API, api.deepseek.com). Text runs server-side on Cloudflare’s edge network.
Photo analysis: Qwen3-VL-Plus
Palm reading and face reading analyze your uploaded photo with Qwen3-VL-Plus (qwen3-vl-plus, Alibaba Cloud DashScope). The photo analysis runs on our VPS through a secure tunnel; the written report is then generated by DeepSeek.
Image generation: Doubao Seedream 5.0 Pro
The AI Image Generator creates images with Doubao Seedream 5.0 Pro (doubao-seedream-5-0-pro-260628) from ByteDance’s Volcengine Ark platform.
Caption vision: LLaVA 1.5 7B
The AI Caption tool understands your image with LLaVA 1.5 7B (@cf/llava-hf/llava-1.5-7b-hf, the open-source LLaVA model running on Cloudflare Workers AI), then writes the captions with DeepSeek.
Speech recognition: Qwen3-ASR
AI subtitles transcribe audio with the open-source Qwen3-ASR model, running on our VPS. This feature is currently paused on the site.
OCR: Tesseract
The image-to-text and OCR tools read text from images and PDFs with Tesseract OCR (open source). The free tools run entirely in your browser; the cloud OCR tools run on our VPS.
How we keep this page current
Models change as providers ship new versions, and we may switch providers. Whenever the model behind a feature changes, we update this page and the model shown on that tool’s page.