Taby's local brain
You don’t pick models, sizes or settings. Taby picks one, sets it up, tests it and runs it on your computer.
Right now that’s Qwen 3.5 4B, an open model from Alibaba. It did best on our test of 114 real jobs. Taby tuned it for everyday jobs like tasks, reminders and notes.
Updated October 10, 2026
Coming soon
We’re training our own small models for what Taby does every day: your tasks, reminders, notes and search.
Done
Picked the best open model
Qwen 3.5 4B won our test of 114 real jobs.
Done
Tuned it for everyday jobs
Tasks, reminders and notes.
Now
Training our own small models
Made for what Taby does every day. The newest one is training right now.
Soon
Showing you how they did
How they’re made, how they score, and which computers they run on. Right here.
How we picked the Brain
We gave small open models the same 114 real jobs, like adding, moving, snoozing and completing tasks. Qwen 3.5 4B got the most right.
Real jobs done right, out of 114
Longer bar = better
| Model | Score of 114 | In plain words |
|---|---|---|
| Qwen 3.5 4B | 94–101 | Best at doing things: moving, snoozing, completing and deleting. No broken actions in any run |
| Gemma 4 E4B (today’s Standard) | 77–98 | Close on a good run, but less steady. 1 to 5 broken actions per run |
| Gemma 4 E2B (today’s Light) | 68–81 | Misses more of the doing. 1 to 5 broken actions per run |
Lowest to highest score over several runs of the same test.
What’s inside Taby
The Brain is the AI model. It writes Taby’s answers and does the job. Listening and talking are two separate parts. All three run on your computer, or Taby Cloud can do them for you.
Three pieces work together
All on your computer
NVIDIA Parakeet turns what you say into words. Hears 8 languages.
Qwen 3.5 4B, tuned by Taby. Thinks, answers and does the job. Reads images too.
Kokoro and Supertonic read the answer out loud. Talks in 11 languages.
Hardware fit
It needs 8 GB of memory or more (a graphics card on Windows). Our free browser test checks your computer in a few seconds. If it can’t run the local AI, Taby Cloud does the AI part instead.
Memory it needs
about 4.3 GB
of graphics memory, so in practice an 8 GB computer
Does it run on mine?
One-time download
Languages
Taby writes its answers in all 14 app languages. On your computer, voice hears 8 of them and talks in 11. Taby Cloud voice covers 100+.
14
languages. Taby writes in every one.
Before the Local Brain
Earlier versions came in four sizes. The update moved everyone to Qwen 3.5 4B in the background, so there’s nothing to pick.
Light
Gemma 4 E2B
3.35 GB model
Standard
Gemma 4 E4B
about 6.1 GB model
Large
Qwen 3.5 9B
5.9 GB model
Extra large
Gemma 4 12B
7 GB model
The Local Brain
Qwen 3.5 4B
about 3.6 GB download, for everyone
B means billions. A 4B model has about 4 billion learned settings inside. Bigger can be smarter, but it also needs a bigger computer.
Size isn’t everything. How a model was trained and packed down matters too. That’s why Taby tests models on real jobs instead of going by size.
Qwen 3.5 4B
Today’s Local Brain
Qwen 3.5 9B
An earlier Brain
Gemma 4 12B
An earlier Brain
Taby’s Brain runs on your computer, so your everyday stuff stays there. For harder jobs, you can turn on Taby Cloud.
Stays on your computer
Nothing you write is used to train models.
Taby Cloud, if you turn it on
For harder jobs. Requests run under Zero Data Retention.
New to this? Local AI, explained in plain words, or see how Taby compares with Ollama and OpenClaw.
Quick answers
One Local Brain for everyone: Qwen 3.5 4B, an open model from Alibaba. Taby downloads it, sets it up and keeps it up to date.
Because it did best on Taby’s own test of 114 real tasks: about 96 of 114, against about 72 for Gemma 4 E2B. It’s mostly better at actually doing things, like moving or completing a task, instead of only answering.
Yes. Taby tuned it for everyday jobs like tasks, reminders and notes.
No, and that’s on purpose: Taby picks, sets up and tests one model so you don’t have to. If you want to try lots of models, apps like Ollama or LM Studio are made for that.
Use Taby Cloud. It runs the AI for you, so it works on smaller laptops and on Windows PCs without a graphics card.
No. The model writes Taby’s answers. Listening and talking use separate speech parts that also run on your computer.
What's next
Use one ready-to-go local assistant, or choose a model file and build the setup yourself
Ready to use
Taby is the all-in-one local option: the model, chat, tasks, notes, planning, and voice work together in one assistant on your computer
See what Taby can doDIY setup
Choose a GGUF model file on Hugging Face, then run it locally with Ollama, LM Studio, or llama.cpp. You manage the runtime and updates
Browse GGUF models