Skip to content
Change language: EN
My learningAdd to Chrome
Track progress 0%

Lesson 1 5 min

Find a model that fits your device

AWFlow runs AI models inside your browser. Find one by task, check that it fits, and install it once.

Three engines, one page

Open Local AI in the sidebar (or Settings › Local models). Once a model is installed it works offline, costs nothing per use, and no image, text or audio leaves your device. You choose a model; AWFlow picks the engine:

  • WebLLM: chat and text-generation models (Llama, Qwen, Phi and more) on WebGPU.
  • TensorFlow.js: vision, text and embedding models.
  • transformers.js: the widest range of tasks, from image captioning and translation to Whisper speech-to-text.
The Local AI page: storage used by on-device models split by engine (WebLLM, TF.js, Transformers), a task list on the left (Chat, Embeddings, vision and audio tasks) and model cards on the right, such as Llama 3.2 1B Instruct with its size, variants and install state.The Local AI page: storage used by on-device models split by engine (WebLLM, TF.js, Transformers), a task list on the left (Chat, Embeddings, vision and audio tasks) and model cards on the right, such as Llama 3.2 1B Instruct with its size, variants and install state.
Tasks

The page groups models by task: Chat, Image classification, Object detection, Speech-to-text, Embeddings and more. Use the search box to filter by name, and the engine filter to narrow the list to one engine.

Pick a variant

Open a model to see its variants: the same model at different precisions. A smaller variant trades a little quality for less memory and a faster load. Each variant shows:

  • an estimated size, marked ≈ where the browser doesn’t expose an exact figure;
  • a device fit: Fits comfortably, Tight fit or Exceeds memory;
  • whether it’s installed.

Models that exceed your device’s estimated memory are still listed and installable, with a warning: they may run slowly or fail to load.

Install, test, keep

Click Install on a variant. Downloads run in the background, one at a time, and you can cancel them from the download panel in the corner. When it’s done, click Test: AWFlow loads the model and runs a quick check. If the test fails, try a smaller variant.

The detail view of Llama 3.2 1B Instruct on the Local AI page, scrolled to Use it in (Web LLM, Transformers Chat, Run Local Model) and Variants: four precisions (q4f16_1 ≈ 879 MB, q4f32_1 ≈ 1.1 GB, q0f16 ≈ 2.5 GB, q0f32 ≈ 5.0 GB), each with an Install button.The detail view of Llama 3.2 1B Instruct on the Local AI page, scrolled to Use it in (Web LLM, Transformers Chat, Run Local Model) and Variants: four precisions (q4f16_1 ≈ 879 MB, q4f32_1 ≈ 1.1 GB, q0f16 ≈ 2.5 GB, q0f32 ≈ 5.0 GB), each with an Install button.
Install a variant

Browsers can delete stored data, models included, when space runs low. Turn on Keep models installed in the storage header; when the browser grants it, the header shows Protected.

Key takeaways

  1. You pick a model; AWFlow picks the engine (WebLLM, TensorFlow.js or transformers.js).
  2. Device fit and the estimated size tell you what your machine can run.
  3. Keep models installed asks the browser not to clear big downloads when space runs low.