a free, open-source Cotypist alternative for Mac

Both apps keep generation on your Mac and place suggestions directly in the text fields you already use. Neither requires copying your writing into a separate chat window. Cotabby is not a cloud wrapper: it runs with Apple's Foundation Models framework or a local GGUF model.

This page is published by the Cotabby team. We link to Cotypist's own documentation for its plans and deliberately include reasons someone might still choose it.

Cotabby

Cotypist

Cotypist details come from its official pricing page. Cotabby details come from the public source repository and current release. Pricing and features can change; report a mismatch to hello@cotabby.app.

Cotabby is free software. There is no account, license key, daily word counter, or paid model tier. The cost of generation is absorbed by hardware you already own because inference stays on your Mac.

Cotypist Free currently includes 100 completed words per day and a 30-day Pro trial. Its annual pricing page lists Plus at $6 per month and Pro at $9 per month when billed annually. Plus removes the daily completion limit for one Mac; Pro adds up to three Macs and its full model and context feature set.

If you only accept an occasional suggestion, Cotypist Free may be enough. If autocomplete becomes part of every email and message, the difference between a daily allowance and unlimited completions is more consequential.

Both products advertise on-device processing. Your writing does not need to be sent to a remote language model merely to produce the next suggestion.

Cotabby adds verifiability: its source, permissions, model integration, and network behavior can be inspected in public. Cotabby does not include telemetry. Network access is used for user-initiated model downloads and update checks. Read the complete Cotabby security model and privacy policy.

Cotabby offers two engines. On macOS 26 or later with compatible hardware, Apple Intelligence uses Apple's built-in Foundation Models runtime. On macOS 14 or later, Cotabby can instead run a local GGUF model through llama.cpp.

Apple Intelligence avoids a separate multi-gigabyte model download. GGUF support provides more control: you can select a model based on latency, quality, memory, and language requirements rather than a subscription tier. See the Apple Intelligence setup guide.

Cotypist instead curates and recommends its model catalog. That is a reasonable tradeoff for users who want fewer decisions and prefer the app to select an appropriate model for their Mac.

There is no honest universal winner. Completion quality depends on the model, available context, language, sentence length, Mac hardware, and how closely a product's defaults match your writing. A screenshot of one successful sentence does not settle that question.

• Try both products in the same three applications.
• Use the same email, chat, and long-form writing tasks.
• Record accepted suggestions rather than suggestions merely shown.
• Watch latency and memory pressure during normal work.
• Keep each product's default settings for the first test.

Cotabby's practical advantage is that you can keep testing models without a paid catalog restriction. Cotypist's practical advantage is a more opinionated default experience.

You can test Cotabby alongside Cotypist without deleting your existing settings. Disable one app globally while evaluating the other so both do not attempt to display suggestions in the same field.

1. 1. Install Cotabby from the latest GitHub release or Homebrew.
2. 2. Grant Accessibility, Input Monitoring, and optional Screen Recording permissions.
3. 3. Select Apple Intelligence or download a local GGUF model.
4. 4. Disable Cotypist temporarily and test Cotabby in your normal apps.
5. 5. Adjust completion length, context, and disabled apps only after testing the defaults.

The full walkthrough is in the Cotypist migration guide.

Ready to evaluate it yourself? Check app compatibility before installing.