Model picker UI: letting users choose an LLM
A good model picker shows a short list with what people actually choose by – speed, quality and cost – instead of model names alone, keeps a sensible default, puts reasoning or effort settings next to the model, and makes fallbacks visible when a model fails. Keep it inside the composer, one click away.
What people choose by
Most people can't tell models apart by name. Show what changes for them: how fast it answers, how good it is for the task, and what it costs – as simple bars or labels, not benchmark tables.
- Model Picker – The model menu of an AI chat – point at a model and its speed, quality and cost bars fill, switch on extended thinking and its effort unfolds, while the tray rolls the first-token time and price
Keep it in the composer
The picker belongs next to the send button, so people switch per message without leaving the conversation. Show the current model's name in the button, and remember the last choice.
- Chat Composer – Prompt composer with model picker, voice input, app connections and approval request
Reasoning and effort
Extended thinking or effort levels change speed and cost a lot. Put them right under the model they belong to, and only show them for models that support them.
Fallbacks and comparing
Models fail and rate-limit. If your product falls back to another model, show it – which model answered and why – instead of hiding it. For tasks where quality matters, running the same prompt on several models and comparing the results helps people pick.
- Fallback Chain – One request finding a model that answers, as a small flow diagram – the request branches out to each model in its fallback order, a light runs to the one being tried, failed branches break red, and the one that answers runs on to the response
- Parallel Attempts – An agent tries the same task several ways at once and the best one wins – every attempt races as a row with its model, approach and live bar, then gets a score; the best is marked and picked, and you can switch to another
Limits and context
Models differ in context size. When a conversation gets long, show how full the context is before it runs out, not after.
- Context Window – What fills the model's context, as a field of cells – hover a part to light it up, remove files, and compact the conversation while its cells shimmer and clear
- Context Map – What fills an agent's context window, as a living treemap – system prompt, tools, memory, every file, the chat and tool results as tiles that glide as it grows, the free space dashed, and compaction squeezing old turns into a summary
Questions
Should I let users pick the model at all?
For power users and developers, yes. For everyone else, choose a good default and offer a simple choice like fast or best – and still show which model answered.
How many models should the picker show?
Three to six covers most products. Group the rest under "More models" so the main list stays scannable.
Updated 2026-10-29