December 2024

# December 5

We have images now 📸 You can upload photos from your device, or just paste it straight to the box. All photos are never stored on our side, only in your device ("IndexedDB" for the devs). They will gone as soon as you delete the conversation.

# December 7

Chat is now available immediately right after load 🐰 Ask Box always take less than 1 second to load, but previously there was a 2-second warming up right after, where typing might be laggy. With multi-threading magic, we moved the warm-up to other place, so now the app is fully interactive as soon as you see it!

This does not only improve the initial experience but also the smoothness of every question! Now all heavy processing is done behind the scene, leaving the main thread fully for maximum animation frames 💨

Thanks @huy for reporting the issue and taking part in the build discussion 🙇

# December 8

Line breaks are back in user messages 🪜 This was a minor but annoying bug for a while, sorry!

Thanks @huy, @chinh, and @hai for reporting the issue 🙇

# December 9

Companions now stay with threads 🥾 When we switch to an existing thread, the thread’s latest companion will be selected automatically, helping us continue the conversation smoothly.

Thanks @my (Pro) for suggesting the idea 🙇

# December 14

We supported Gemini 2 Flash (Google) and Llama 3.3 70B (Meta) 🌟 Both of them outperform their larger predecessors, Gemini 1.5 Pro and Llama 3.1 405B respectively, while being much faster as well.

# December 22

Gemini Thinking Mode is in the house 🧠 Trained to generate the "thinking process", Thinking Mode is capable of stronger reasoning capabilities than the base Gemini 2.0 Flash model.

# December 24

Companion Model is more visible 🔦 Companions now have their model right next to their name in important places, such as the ask box. Moreover, if you're fan of one companion per model, we now support leaving the "name" field empty to automatically use its model's short name.

Thanks @my (Pro) and @minh for suggesting the idea 🙇