Hey Anna community! 
Happy Monday! We’ve just rolled out one of our most exciting updates yet — 6 brand-new models landed on the platform, plus a refreshed model picker and live, always-accurate plan info. Whether you’re chatting with Anna every day or building the next great Anna App, there’s something here for you. Let’s dive in! 
New Models — The Stars of This Release 
Qwen International family (new provider!)
We’ve onboarded Qwen International (Alibaba Cloud Model Studio, Singapore region) as a brand-new provider, bringing three powerhouse models:
| Model | Highlights | Plans |
|---|---|---|
 Qwen3.6 Flash |
Multimodal (image  + video  ), 1M context, 64K output — blazing fast & super affordable at ×1.99 |
 Free / Pro / Max |
 Qwen3.7 Plus |
Multimodal, 1M context, 256K thinking chain  , great quality-to-cost balance at ×3.31 |
Free / Pro / Max |
 Qwen3.7 Max |
The flagship — top-tier reasoning & writing, 1M context, ×15.5 |  Pro / Max |
Yes, you read that right — Free users get Qwen3.6 Flash AND Qwen3.7 Plus, both with a full 1M context window! 
New via OpenRouter
| Model | Highlights | Plans |
|---|---|---|
 Kimi K2.6 |
Moonshot’s latest — vision support  , 262K context, ×6.78 |
Pro / Max |
 MiniMax M3 |
Vision + video understanding, 1M context, incredible value at ×2.4 | Free / Pro / Max |
 Gemini 3.5 Flash |
Google’s newest Flash — 1M context, audio support  , ×18.0 |
Pro / Max |
A Cleaner, Friendlier Model Picker
The Model Selection area on the /llm page got a glow-up! 
Models are now grouped by family (Qwen, Gemini, Kimi, MiniMax, MiMo…), each family on its own neat row
Cards are simplified to show just what matters: name, provider, max tokens, and cost multiplier
Hover any card to see its full description
Search, sort, and type filters all still work exactly as before
Finding your perfect model has never been faster! 
Pricing Page Now Shows Live Model Counts
The “Available Models” number on /pricing is no longer a static label — it’s now computed in real time from the actual model catalog. 
Every time we add a new model (like today! 
), your plan’s model count updates automatically. What you see is exactly what you get — no stale numbers, ever. 
For Anna App Developers
All new models are available through the platform’s sampling & agent APIs right away — agent.session.create, LLM sampling from your Executa plugins, the works. 
Try the multimodal models (Qwen3.6 Flash, MiniMax M3) for vision-powered apps
Low multipliers = more experiments per Energy point
Pair them with the recent structured output emulation in the Anna AI CLI for fully local rehearsals
How to Try Them
- Head to Settings → LLM (
/llm)
- Pick your new favorite from the freshly grouped model list
- Start chatting / building!
We’d Love to Hear From You!
Which model is your new daily driver? Built something cool with the 1M context window? Found a killer use case for video understanding? Drop a comment below — your feedback directly shapes what we ship next.
Happy chatting & happy building! 
— The Anna Team
They were quietly some of the biggest stars of this release, so let me give them the spotlight they totally deserve 
Xiaomi MiMo family — surprise stars of the show! 
), tackles tasks that take human experts days with 1000+ tool calls. 1M context · 131K output · ×1.74
MiMo-V2.5
And MiMo-V2.5-Pro is now one of your default Starter partner models, so you might’ve already made friends with it 
