Kerlig Alternative for Mac: 350 Models or One Perfect Fix?

What is Kerlig?

Kerlig is an AI writing assistant for macOS that has earned a solid reputation in the Mac community, and it deserves real credit. You press a global hotkey (Option+Space by default), and a floating window appears where you can fix grammar, rewrite text, compose replies, summarize, or chat with documents, web pages, and ebooks. It connects to 350+ AI models through your own API keys — OpenAI, Anthropic, Google, Groq, and OpenRouter — and also supports local models via Ollama, LM Studio, and Apple Foundation Models. It handles attachments and images, shows reasoning traces for thinking models, and keeps a chat history. It runs on macOS 12 and later, on both Apple Silicon and Intel Macs.

Pricing is one-time rather than subscription-based. As of this writing, kerlig.com lists a Basic license at $49 (discounted from a $247 regular price) for one Mac, a Pro license at $79 for two Macs, and a Team license at $297 for ten seats. Licenses include a year of app updates, with promotions occasionally bundling lifetime updates — check the current terms before buying. On top of the license, you pay your AI providers directly for the tokens you use with cloud models.

So why would anyone look for a Kerlig alternative? Because Kerlig has grown into something close to a full LLM client. If what you actually want is a small, fast tool that fixes your writing where you type, that breadth becomes overhead.

Why look for a Kerlig alternative?

None of these points make Kerlig a bad product. They are trade-offs that come with its scope:

• Breadth you may never use. Document chat, web page capture, ebook reading, vision, reasoning traces, headless mode — Kerlig is a serious multi-purpose AI workbench. If your daily need is “fix this sentence before I hit send,” most of that surface area sits unused. Even users who like Kerlig note that it lives in an in-between zone: broader than a grammar fixer, but not quite a full replacement for a dedicated chat client either.
• A window-based workflow. Kerlig works through its own floating panel: select text, summon the window, pick an action, review the output, then insert it back. That is flexible, but it is more steps than having corrected text simply appear where your cursor already is.
• Layered costs. The license is one-time, which is great, but tiers run up to $297, and cloud usage is billed separately by your API provider. Light users end up managing API keys and token billing for what is often a one-shortcut task.
• Setup friction. To get the headline 350-model support, you bring your own API keys and configure providers. That is a feature for power users and a chore for everyone else.
• Update policy fine print. Standard licenses include one year of updates; what happens after that depends on the deal you bought under. It is worth reading closely, as promotional terms have varied.

If you recognize yourself in these points, a focused Kerlig alternative makes sense — something built around one job that does it with as little ceremony as possible.

The focused alternative: WunderType

WunderType starts from the opposite premise. Instead of a do-everything AI panel, it does one thing: it fixes and transforms text in place, in any Mac app. Select text in Mail, Notes, Slack, Chrome, or VS Code, press a keyboard shortcut, and the corrected text replaces your selection right where it is — via the macOS Accessibility API, with no floating window, no copy-paste round trip, and no chat session.

It ships with five built-in modes — Correct Grammar, Improve Writing, Make Concise, Make Formal, and Make Casual — plus unlimited custom prompts, each with its own shortcut. So “translate to German” or “rewrite as a friendly Slack reply” is one keystroke, not a conversation.

You still get real AI choice, just a curated one. WunderType supports four providers: On-Device AI powered by Apple MLX local models (nothing leaves your Mac), Ollama for local models you already run, OpenAI with your own API key stored in the macOS Keychain and sent directly to the API, and OpenRouter for access to a wide range of hosted models. There is no WunderType backend server in the middle — ever.

And privacy is not a settings toggle, it is the architecture: zero analytics, zero telemetry, no account, no clipboard monitoring, and the app runs fully sandboxed. If that is the main reason you are shopping around, our overview of what a private AI proofreader should look like goes deeper.

Kerlig vs. WunderType at a glance

	Kerlig	WunderType
Approach	Floating AI window: actions, document chat, attachments, history	In-place text correction via keyboard shortcut, no window
Model support	350+ models via your API keys (OpenAI, Anthropic, Google, Groq, OpenRouter) plus Ollama, LM Studio, Apple Foundation Models	Four providers: On-Device AI (Apple MLX), Ollama, OpenAI (own key), OpenRouter
Pricing	One-time license; as of this writing $49–$297 depending on tier, plus provider token costs; one year of updates	One-time purchase on the Mac App Store, no subscription; local AI costs nothing to run
Privacy	Local model options available; cloud usage goes to your chosen provider	Zero analytics or telemetry, no account, no clipboard monitoring, App Sandbox, no backend server; fully offline with On-Device AI
Footprint	Full-featured assistant app	Native Swift/SwiftUI, under 5 MB of memory
Platform	macOS 12+, Apple Silicon and Intel	macOS 15+, optimized for Apple Silicon

The honest summary: Kerlig wins on raw capability and model count. WunderType wins on speed of workflow, privacy guarantees, and simplicity. Which one is “better” depends entirely on whether you want an AI workbench or a writing reflex.

Is Kerlig a subscription or a one-time purchase?

Kerlig is a one-time purchase — no subscription. You buy a license and then pay only for the AI tokens you use with your own cloud API keys, or nothing extra if you run local models. The nuance is the update window: standard licenses include one year of app updates, and lifetime-update offers appear during promotions. WunderType is also a one-time purchase, sold through the Mac App Store, with no separate token billing when you use On-Device AI or Ollama. If subscription fatigue is what brought you here, both apps respect that — the difference is how many moving parts you take on afterwards. We covered a similar trade-off in our RewriteBar alternative comparison, where lightweight menu-bar tools sit between the two extremes.

Can you keep everything on-device?

With both apps, yes — but the defaults differ. Kerlig supports local models through Ollama, LM Studio, and Apple Foundation Models, alongside its large cloud catalog. WunderType makes local-first the headline: its On-Device AI runs Apple MLX models directly on your Mac, so grammar correction works with no internet connection, no API key, and no per-token cost. For most people, that is the practical difference between “privacy is possible if you configure it” and “privacy is the default.” If you are weighing on-device options more broadly, see how WunderType compares to Apple’s built-in Writing Tools, and why a dedicated local tool beats a Grammarly-style cloud service for sensitive text.

Verdict: workbench or reflex?

Choose Kerlig if you genuinely need the breadth: chatting with PDFs and web pages, switching between hundreds of models, attaching images, composing long replies in a dedicated window. It is a polished, actively developed app, and for power users the $49 entry price is fair.

Choose WunderType as your Kerlig alternative if your real job is writing well in the apps you already use. One shortcut, corrected text in place, five modes plus your own custom prompts, local AI by default, and a privacy story with no asterisks: no analytics, no account, no server between you and the model. It is the difference between opening a tool and having a reflex.

Download WunderType from the Mac App Store — one-time purchase, no subscription, and your words never leave your Mac unless you choose a cloud provider.