iiinova — where your digital self keeps evolving

KerligAI @kerlig.com · 22/09/2025

Exciting update! 🍏 Kerlig 2.5.0 now supports Apple Foundation Model (on-device) on macOS 26! Use your favorite Kerlig actions with the same model used by Apple Intelligence!

Apple Foundation Model supported by Kerlig app

110

KerligAI @kerlig.com · 19/08/2025

New on the Blog! 🤓 The 10-Minute Market Research Method That Actually Works

100

KerligAI @kerlig.com · 05/08/2025

🚀 Kerlig v2.4.0 now integrates with Cerebras - the world's fastest AI inference platform! ⚡ Lightning-fast speeds 🆓 Generous free tier 🤖 Access to Llama 4 Scout, Qwen 3 Coder 480B, and more cutting-edge models Making AI workflows more efficient than ever! #AI #MachineLearning #Kerlig

001

KerligAI @kerlig.com · 19/05/2025

Fixes: → Resolved an issue where capturing large amounts of content was sometimes unsuccessful → Fixed occasional window flickering in dark mode when opening → UI improvements

010

KerligAI @kerlig.com · 19/05/2025

Kerlig 2.2.0 release! ⚡️ → Speed up your workflow even more by using keyboard shortcuts for Insert in... V, Copy to clipboard C and Regenerate R → When using OpenRouter, the app name (Kerlig) is now displayed on your OpenRouter dashboard → Update for Gemini 2.5 Pro model (from 03-25 to 05-06)

110

KerligAI @kerlig.com · 14/05/2025

Mathematical expressions are now supported in Kerlig 2.1.0 ⚡️

110

KerligAI @kerlig.com · 12/05/2025

- Allow attaching .epub exported from Apple Books (unpacked .epub) - Fix for loading webpages, where in some cases it could result in empty content - Hide UI artifact showing up for some users Enjoy!

000

KerligAI @kerlig.com · 12/05/2025

Kerlig 2.0.1 ⚡️ is now released with a few updates based on feedback, following the major 2.0 release. - Return focus to the app from which Kerlig was launched after closing the Kerlig window with the Escape key - Remove Presets content length limit (was 10,000 characters)

100

KerligAI @kerlig.com · 07/05/2025

- Menu bar now shows a "Renew" button when app updates have expired. - Correct Preset and Model are now displayed in the header while streaming response. - Improved performance when loading webpages (as attachments or citations). Thank you for choosing Kerlig!

000

KerligAI @kerlig.com · 07/05/2025

🔧 Improvements - The selected text is always used by default, even if the same text is on the clipboard. - Summarize, Explain and Translate actions now work with webpages and files.- All keyboard shortcuts are consistent with the current keyboard layout.

100

KerligAI @kerlig.com · 07/05/2025

🧠 New models - Updated lineup of models for OpenAI, Gemini, and Groq with the latest ones.

100

KerligAI @kerlig.com · 07/05/2025

💻 Window management - There is a new option "Always on top" in the menu bar (turned on by default), that replaces the "Pin window" which is now removed. - The window position is now saved between sessions, and you can reset it using the "Center on Screen" option in the menu bar.

100

KerligAI @kerlig.com · 07/05/2025

You can also export and share them with others! The new Headless mode allows you to run actions in the background, editing selected text in-place or saving the output to the clipboard.

100

KerligAI @kerlig.com · 07/05/2025

Now with all internal prompts editable and open-sourced, you can customize them to perfectly fit your personal workflows. Moreover it’s now possible to create new actions, assign shortcuts, make them use selected text, clipboard content, files, or webpages.

100

KerligAI @kerlig.com · 07/05/2025

Kerlig 2.0 is here with Custom Actions ⚡️⚡️⚡️

110

KerligAI @kerlig.com · 12/04/2025

The art and the science. The thinking and the doing✌️

020

KerligAI @kerlig.com · 10/04/2025

Love these messages! ✨

000

KerligAI @kerlig.com · 04/04/2025

You can add custom models, such as Gemini 2.5, to Kerlig at any time. Use the name "gemini-2.5-pro-exp-03-25" when adding. Gemini 2.5 Pro excels in advanced reasoning, complex problem-solving, coding, and web app development. ✨ Here is a quick demo:

010

KerligAI @kerlig.com · 25/03/2025

WIP → Custom Actions! ✨ Need your feedback – what would you like to add/remove here?

000

KerligAI @kerlig.com · 24/03/2025

New on the Blog → Guide to Perplexity Deep Research with examples 🧠 Includes SaaS market research, travel planning, quantum computing...

120

KerligAI @kerlig.com · 18/03/2025

🚀 Thrilled to announce Perplexity is now a first-class integration in Kerlig 1.8.0! ✨ Web search with citations 💡 Transparent reasoning ⌨️ CLI support for Team License owners Plus new keyboard shortcuts for model & preset selection, better streaming, and support for macOS 10.15+

110

KerligAI @kerlig.com · 10/03/2025

Quality-of-life improvements: Added Project/Organization ID fields to OpenAI settings, edit preset names in the sidebar, and settings window now appears in the dock.

100

KerligAI @kerlig.com · 10/03/2025

Model management got a major upgrade: rename integrations, add custom models, edit modalities, search models, and a fresh UI design.

100

KerligAI @kerlig.com · 10/03/2025

Added Anthropic via Amazon Bedrock and the latest models: Claude 3.7 Sonnet, GPT-4.5 Preview, and o3-mini.

100

KerligAI @kerlig.com · 10/03/2025

Added support for OpenAI-compatible services including Mistral, xAI, Perplexity, DeepSeek, LM Studio and any custom endpoint.

100

KerligAI @kerlig.com · 10/03/2025

You can now connect multiple instances of each integration - for example when you are a member of multiple OpenAI Orgs or Projects, or want to connect to multiple Ollama servers simultaneously.

100

KerligAI @kerlig.com · 10/03/2025

Excited to announce Kerlig 1.7.0 with revamped AI integrations and model management!

110

KerligAI @kerlig.com · 01/03/2025

New on the Blog → Prompt Library for DeepSeek models ✨ Includes blog writing, proofreading, marketing, and more.

120

KerligAI @kerlig.com · 15/02/2025

The initial model load time is a few seconds, and then it generates text at a decent speed, this took approximately 2 seconds. GPU memory is, as expected, maxing out when generating.

000

KerligAI @kerlig.com · 15/02/2025

Taking local model "Mistral Small 24b" via Ollama for a spin! Overall, it has decent speed and accuracy! 👏 Sometimes it adds additional text even though the "fix spelling" action has an explicit prompt saying not to add anything. Tested on M1 Max (32GB).

110

KerligAI @kerlig.com · 14/02/2025

If you want to fix spelling & grammar with local models, I found the best results using Gemma 9b and Llama 3.2 11b models. Smaller models can get confused easily, while models with more than 14b are starting to be quite slow, so 9b and 11b are kind of a sweet spot.

010

KerligAI @kerlig.com · 11/02/2025

Awesome to hear that! 😊

020

KerligAI @kerlig.com · 10/02/2025

The Llama 3.2 Vision model has been added to our default model list – just click the toggle, and it will be downloaded automatically. Another great vision model is LLaVA, which you can add manually by typing the name llava:7b, llava:13b, or any of the 98 total available versions.

100

KerligAI @kerlig.com · 10/02/2025

Kerlig 1.6.2 🚀 adds support for sending images to Ollama vision models, which allows on-device document analysis, extracting text, understanding charts/diagrams, and creating descriptions for visual content.

110

KerligAI @kerlig.com · 08/02/2025

🚀 New on the Blog → Running Gemini 2.0 Models on macOS with 2M token context, 90% cost savings vs competitors, and mind-blowing performance. Learn how to run these powerhouse models right on your Mac with @Kerlig.com in our latest guide 🧵

220

KerligAI @kerlig.com · 08/02/2025

🚀 Kerlig 1.6.1 release! Check out the latest update featuring new Google Gemini models: 1️⃣ Gemini 2.0 Pro 2️⃣ Gemini 2.0 Flash 3️⃣ Gemini 2.0 Flash-Lite 4️⃣ Gemini 2.0 Flash Thinking Plus, Kerlig menu bar icon now alerts you when update is available & bug fixes!

130

KerligAI @kerlig.com · 06/02/2025

🚀 New release: Kerlig™ 1.6.0! Added support for remote team deployments with CLI commands for license & API key management. New default models added to Ollama: DeepSeek R1, Llama 3.1, Qwen 2.5, Phi-4. Plus, fixed Anthropic prompt caching issue. Enjoy!

120

KerligAI @kerlig.com · 29/01/2025

You can run full DeepSeek R1 *671B* in Kerlig via @openrouter.bsky.social, plus its 4 other versions 👇

110