Modes & Models
Two dials shape how Zinley answers: the mode (what kind of work this is) and the model (how much horsepower it gets). Most of the time you can leave both on Auto.
Modes
A mode tells Zinley what shape the job is — a quick lookup, a deep research dive, a spreadsheet. It changes how long Zinley works, which tools it reaches for, and what the answer looks like when it lands.
| Mode | Use it for |
|---|---|
| Auto | Everything, unless you know better. Zinley reads your message and picks the mode itself. |
| Quick Search | A fast factual answer from the web — dates, prices, "is this still true?" |
| Research | A thorough dive across many sources, ending in a written brief with sources. |
| Analysis | Reasoning over material you've given it — documents, numbers, options, trade-offs. |
| Deep work | Long multi-step jobs that need planning and several passes to finish. |
| Documents | Drafting and editing long-form writing — reports, proposals, letters. |
| Sheets | Building and reworking spreadsheets and tabular data. |
| Design | Visual work — layouts, mockups, and image-led output. |
Auto mode
Auto is the default and it's the right answer most of the time. Zinley reads what you asked for and routes it — a one-line question doesn't get the deep-research treatment, and a "look into this properly" doesn't get a two-sentence reply.
Reach for a specific mode when you want to override that judgement: force a deep dive on something that looks simple, or force a quick answer on something that looks big.
Models
The selector at the bottom-left of the message box controls which models do the work.
Zinley doesn't run on a single model. A boss model reads the task and directs it, doing the high-value thinking itself and delegating the rest to a team of workers that runs in parallel — mostly on cheaper models, escalating to a premium one only when the task actually needs it.
What each plan unlocks
| Plan | Model picker |
|---|---|
| Free | No picker — everything runs on Auto. Zinley chooses for you, every time. |
| Plus | Picker unlocked, plus premium models. |
| Pro | Adds more advanced models on top of the Plus set. |
| Max | Everything — the full lineup, no model held back. |
Being on Free doesn't mean weaker answers for everyday work — Auto still routes each task to a capable model. What you're unlocking with a paid plan is the ability to insist on a particular one.
Models cost different amounts
Every model carries a credit weight, and your message is charged accordingly:
| Weight | Meaning |
|---|---|
| 0.5× | Light and fast — costs half a normal message |
| 1× | Standard rate |
| 2× | Premium reasoning — costs double |
So a long conversation on a 2× model burns through credits four times faster than the same conversation on a 0.5× one. If you're watching your balance, leaving it on Auto is the cheapest sensible default — it only spends the expensive models where they earn it.
Speed & Model — the model for email, calls, and texts
The picker in the message box only exists in chat. Email, phone calls, and texting have no picker in them — so the model they use is the one you set on the Speed & Model page, in the sidebar under Configure, directly below Your Zinley.
Open the Model card and pick one:
| Choice | What it means |
|---|---|
| Auto (default) | Picks the best model for each prompt, balancing capability, speed, and cost per request. |
| A named model | Every request runs on that one model — including the channels with no picker. |
The card lists a few models up front — at the time of writing Claude Opus 5 (newest Opus — frontier coding, agents & computer use), Claude Fable 5 (most capable — deep, always-on reasoning) and GPT-5.6 Sol (flagship — hardest coding & advanced agents) — and Show all models opens the full lineup. Which models you can choose from depends on your plan; the list itself changes as new models ship, so treat the names above as an example of the shape rather than a fixed menu.
This one choice applies to chat, email, and calls, and to Telegram, Discord and iMessage & SMS — everywhere that has no picker of its own. Pinning a premium model here means your agent answers every email and every call on it, at that model's credit weight. Leave it on Auto unless you have a reason not to.
The same page holds the Image & video card — the models behind image, video, and music generation, covered in Creative work.
Speed & intelligence upgrades
Further down Speed & Model are the optional premium upgrades to how Zinley runs:
- Priority processing — routes supported requests through a faster lane, so replies feel snappier on busy, multi-step tasks. You're only charged the premium when priority is actually delivered — never for standard responses.
- Boss model — choose how smart the model that plans and directs the work is. The standard boss is included with every paid plan; deepest-reasoning bosses cost more per request but plan harder problems better.
- Experimental deep-work models — opt in to a newer model for deep work. If it errors, your task falls back to a proven model automatically, so the work always finishes.
Turning any of these on only changes how the work is done — the team of models still runs as usual underneath.
Which should I change?
- Answer feels too shallow? Switch the mode to Research or Deep work — that matters more than the model.
- Email or calls answering with the wrong depth? They don't read the chat picker — change the model on Speed & Model.
- Burning credits too fast? Go back to Auto and let Zinley route.
- Hard reasoning problem? A premium model or a deeper boss is worth it here.
- Just need a fact? Quick Search on Auto is the cheapest, fastest path.
Related
- Chat Basics — sending messages, attachments, and steering.
- Billing & Usage — credits, plans, and top-ups.
- Plan Limits — what each plan includes, side by side.
- Creative work — images, video, and music generation.