Choosing a model

StickyPrompts puts the leading models from OpenAI, Anthropic, Google, xAI, Moonshot, DeepSeek, Meta, Mistral, Alibaba and others behind one composer. You can switch model at any point in a chat, so you are never locked into one provider’s strengths and weaknesses. This page shows the three ways to choose and how to read the model selector.

Three ways to choose

Select the model button in the composer (it shows the current model’s name, for example Claude Opus 5.5). A menu opens with three sections.

app.stickyprompts.com/conversations/0VxTk5Pc 1 Let StickyPrompts pick 2 Your recent models 3 Every model

Auto select

Pick a goal and StickyPrompts chooses the model for you, message by message:

Mode	Use it when
Best quality	The answer matters more than the wait: analysis, important writing, complex reasoning.
Balanced	Everyday work where you want good answers without waiting long.
Fastest	Quick rewrites, short answers, brainstorming, anything where speed is the point.

StickyPrompts picks a model that suits the request, so you don’t have to keep track of which model is best at what this month.

Recently used

The models you picked most recently, one click away. Handy when you have a favourite for a certain kind of work.

Choose model

Opens the full Select AI Model dialog with every model you are allowed to use.

The model selector

app.stickyprompts.com 1 Filter the list 2 Search by name 3 Credit multiplier 4 Compare instead

Narrow the list (1, 2). Filter by maker with All providers, by what a model can do with All tools, and by Any tags. Switch on Flagship only to see each maker’s top models, or Show legacy to include older versions. Or type in Search AI models… to jump straight to one.

Read a card. Each card tells you:

• Name and maker, for example Claude Opus 5.5 by Anthropic or GPT-6 Astra by OpenAI.
• A description of what the model is built for, sometimes with a Best for line. Select Show more for the full text.
• Knowledge cutoff: the date its training data ends. For anything more recent, use a model with web search.
• Capability tags, such as web_search, web_fetch, file_upload, code_execution, image_generation, file_generation and remote_mcp. They tell you what the model can do inside a chat.
• Badges: Flagship for a maker’s top model, and DEFAULT on your default model.
• The credit multiplier (3) and an estimate of credits per message.

Select a card, then Select Model. To send one prompt to several models at once, switch on Comparison Mode (4) instead. See Compare models.

Credit multipliers

Usage in StickyPrompts is measured in credits. Each model shows a multiplier that tells you how quickly it uses credits compared with other models, plus an estimate of the credits a typical message costs. For example, Claude Opus 5.5 shows 2.133x and about 22.40 credits per message, and GPT-6 Sol shows 1.067x. The actual amount depends on how long your prompt, attachments and the reply are.

Use the multiplier to match the model to the job: a flagship model earns its multiplier on a board summary or a tricky analysis, while a quick reformat is done just as well, and sooner, by a fast model. Your credits used this month are shown in the usage meter at the bottom of the sidebar. Admins see the full breakdown in Usage, credits and billing.

Set your preferred model

The model new chats start with is your Preferred AI model. Change it in Settings > Preferences, together with your preferred response style. Besides every model you are allowed to use, the list offers Auto (Best Quality), Auto (Balanced) and Auto (Fastest), which let StickyPrompts pick the model for each chat.

Model selection behavior decides what happens when you pick a different model in a chat. By default (Always ask) you are asked whether to make it your new default; with Never set selected model as default, switching model in one chat leaves your preferred model unchanged. See Preferences.

Which kind of model for which job

There is no single best model; there is the best model for the job. Some general guidance:

The job	Reach for
Hard reasoning, long documents, multi-step analysis, agent work with many tools	A Flagship model with High reasoning effort, or auto-select Best quality.
Everyday writing, summaries, email replies	Auto-select Balanced, or your preferred model.
Short, repetitive tasks: rewording, classifying, extracting fields	A fast model, or auto-select Fastest.
Anything about recent events	A model tagged web_search, so answers come with live, sourced results.
Working with files you attach	A model tagged file_upload.
Images, video, music and speech	The dedicated apps, which list the models built for each medium.

The fastest way to settle a choice for work you do often is to test it: run your real prompt through two or three candidates with Compare models and keep the one whose answers you would actually send.

Related pages