Love it but…
One thing that I really struggle with is tracking how much context I have used up. If you could add a bar like Silly Tavern has or some other way of tracking how much context is left, that would really go a long way.
Update
Response from developer
Editing AI replies is supported: long-press the AI's message bubble to open the actions menu, then choose Edit. We agree it is hard to discover, and we are working on making the message actions easier to reach. If long-press does not open the menu on your device, please tell us your app build and device at https://github.com/miaoxworld/NativeTavern/issues and we will investigate.
Need better config for TTS
Response from developer
Thanks for the detailed report. Right now the TTS voice list follows your device's system voices, and when you point NativeTavern at an OpenAI-compatible TTS server (for example Alltalk) the voice name has to match a voice that server actually exposes. A dedicated voice dropdown and an explicit OAI default voice are on our list, and we are looking into the Alltalk value error. If you can share the endpoint shape and the exact error text (never include API keys), please open an issue at https://github.com/miaoxworld/NativeTavern/issues so we can reproduce it.
Great options and tons of features
1. Reasoning/thinking doesn’t seem to display for me? I thought I heard you say it does for reasoning supported models but I’ve tried Kimi K2.5, MiMo V2.5 Pro, DeepSeek V4 Pro and Flash and don’t have any thinking shown (yes I have it enabled and I’ve tried auto, minimal, medium, maximum; no difference). I know it’s reasoning because it takes alot longer than if it’s turned off, it just doesn’t show up anywhere. Maybe a bug? I rely a lot on reading the reasoning to tweak my bots and my system prompts.
2. I use Openrouter exclusively. I really like that you can change a AI model in chat at the top left by the character’s name, however it’d be nice if I can fine tune which inference provider (like you can in the ‘LLM Connection’ setting). If I switch the model while in a chat, it just reverts to ‘auto’ for provider, which is bad because certain providers aren’t good and I like to stick to one provider so I can use cache. So then I have to go out of the chat and switch it back to my preferred provider. It would also help a lot to be able to ‘star’ my favorite models so they stay at the top of the list. It’d be nice to switch reasoning effort from here as well, but that’s not as important.
2.1. Big problem with the inference provider option. With Kimi and MiMo the provider is picked correctly. However, with GLM and DeepSeek Models (GLM 4.7 and DSV4 Pro, DSV4.1 Flash tested), the router reverted to AUTO, and DID NOT keep the correct provider I selected. Big big problem.
3. You have to be very careful to hold down on a blank space to the far right of the AI’s chat bubbles to see the option list come up (edit, regenerate, delete, etc.). I didn’t realize you could do this except by accident and it’s pretty hard to do, maybe there is a better way? It almost always only works if I hold down on the very far right edge of the chatbubble and that’s not very reliable or easy.
4. I’d love to be able to rename my chats in the “chats” section and that show up when I click on a character, that way I know which is which at a glance without having to go through the chat’s text and try and remember what roleplay this particular chat is.
5. I have lots of personas and change/restart chats often. It’s kind of annoying having to go over to settings every time I want to change each persona. It’d be nice if, when I start a new chat session, my list of personas will pop up and I can just pick the one I want to use and then start the chat session.
6. The text size is kind of big for me when I’m typing and the max area shown to type it in is quite small. It only holds about 20 words before you have to scroll. It’d be nice to be able to resize it so it fits about 70-80 words before I have to scroll. That lets me re-read my draft before I hit send a lot faster without having to scroll constantly.
7. You have an option that lets the AI write for you, which is great. It’d be nice to also have the option to have it read a draft you wrote and then improve it/lengthen it like some other apps allow, if possible.
8. I’d love to be able to quickly start a new chat with the same character from the three dot menu in the top right corner while I’m already in a chat (just saves a lot of steps).
9. It’d be great to be able to tap at a certain point in a chat (like the long press in 3.) and fork the conversation into a brand new chat at that particular point in the old chat.
10. I would love for folders to be added to the character tab so I can organize characters better.
Response from developer
Thanks for the thorough write-up, it is genuinely useful. We are tracking both problems you hit: reasoning/thinking content not being displayed for some models, and the OpenRouter upstream provider resetting to auto after switching models inside a chat. We have also noted the rest: reaching the message actions (the long-press target is too small), renaming chats, a persona picker when starting a chat, and a taller composer. Thanks for the 5 stars and for spelling it all out.
