ChatWise pricing and API cost: what the licence buys and what the tokens cost

ChatWise costs $0 on its free tier and $29 as a one-time Pro licence — promotional against a $49 list price on September 19, 2026 — and neither price includes a single model token. The vendor says so itself: its pricing FAQ answers "How do I access GPT-4, Claude and Gemini?" with "You bring your own API key, and ChatWise will use it to access the model's API. No subscription required." So the licence is a one-off, and the only recurring number in a ChatWise budget is the API bill you arrange separately.

One disambiguation first, because the search results mix four products. This page is only about the desktop chat client at chatwise.app — macOS, Windows and Linux, made by a single developer — its own pricing FAQ names Kevin as the author. A dating-reply assistant at chatwise.ai, a B2B customer-support chatbot SaaS at chatwiseai.co, and mobile apps from unrelated developers trade under the same name at their own domains — checked September 21, 2026 only far enough to confirm they are separate products; their plans were not read for this page and none of their prices apply here. The desktop client has no mobile app at all, which its own FAQ confirms in the words "There's no plan to release a mobile app at the moment".

A second trap catches automated answers rather than shoppers. The public repository egoist/chatwise-releases is a releases-and-issues tracker that carries no licence file and holds release-sync scripts rather than the application's source, and its newest GitHub Release is v0.10.8 from March 19, 2026, with the last push on July 1, 2026 (GitHub API, September 19, 2026). The shipping product is on a different line entirely: the homepage download button reads Download 26.9.0 once the page hydrates, and a HTTP check on the 26.9.0 Apple Silicon disk image returns 200 with a Last-Modified of September 8, 2026. ChatWise's migration doc explains the break: v26 is the new desktop release line, the app was rewritten in Electron starting in 26.3.0, and releases now use a YEAR.MONTH.BUILD_ID format. Any page or answer built on that GitHub feed is describing a product six months and one rewrite out of date.

ChatWise pricing: what the software costs

All of the above read from chatwise.app/pricing on September 19, 2026. The $29 sits under a banner reading "Discounts ending soon!", so quote it with the date and confirm it in checkout; whether that price is regional, time-limited or tested was not established. Pro unlocks chat history search, custom assistants, web search, artifacts, unlimited MCP tools, early access, one year of updates and priority support. Note one unresolved inconsistency: the pricing page lists web search as a Pro unlock while the homepage lists it among core features, and neither page yields to the other.

Two absences are worth budgeting for, in the good sense. ChatWise stores data locally and lists cloud sync as planned rather than shipped, and its FAQ says there is no plan for a mobile app at the moment. Neither page prices a cloud seat or a second device, which is what follows from having no sync to sell — and it cuts the other way too: a second machine is a second install with its own history.

Three ways to get models in, and what each one bills

This is the separation the licence page does not make for you. ChatWise's homepage names two paths in its own headings — "Bring Your Own API Key" and "Use Your Existing Subscriptions", the latter badged GitHub Copilot and Claude Pro/Max — and a third, running a local model, gets its own provider doc.

Plan prices read from GitHub's Copilot plans page and Anthropic's pricing page on September 19, 2026; the ChatWise behaviour from its Copilot doc.

The Claude row needs a warning rather than a recommendation. ChatWise's homepage advertises "Use Your Existing Subscriptions" and names Claude Pro/Max. Anthropic's legal and compliance page says OAuth authentication "is intended exclusively" for subscription purchasers using Claude Code and other native Anthropic applications, that Anthropic "does not permit third-party developers to offer Claude.ai login into their own applications, or to route requests through Free, Pro, or Max plan credentials on behalf of their users", and that it "may do so without prior notice" when enforcing. Both were checked on September 19, 2026, and this page does not claim the feature currently works or currently fails — it was not tested here. What is safe to say is that a route whose availability rests on another company's enforcement discretion is a poor foundation for a year's budget, while a key billed to you is not exposed to that risk at all.

The ChatWise features that spend tokens you did not type

This is where a chat-client API bill actually differs from a raw API bill, and none of it appears on a pricing page.

Read from summarize, auxiliary visual, auxiliary think, documents, agent mode and web search on September 19, 2026. The web-search row is the cheerful one: in a client where search is often assumed to be a metered add-on, two engines here cost nothing at all. The @think row is the expensive one: the auxiliary-think doc has the reasoning model produce a chain and the main model consume it, so those tokens are paid for once on the way out and again as input on the main call. That is an inference from the documented flow, not a cost ChatWise publishes.

A worked ChatWise API cost estimate

These are illustrative token arithmetic, not measured task costs and not a bill ceiling. Assume a month of desktop chat at 20 working days, 10 conversations a day, 6 turns each — 1,200 turns — averaging 5,000 input tokens per turn as the thread accumulates and 500 output tokens per turn. The title column assumes ChatWise generates one title per conversation from roughly 1,500 tokens of the opening exchange, billed to the same key. Rates are live Kunavo catalog prices per million tokens.

The reading that matters for a buyer: under these assumptions the model choice moves the twelve-month figure from roughly $44.74 to $223.68, while the licence is a single $29 paid once. Do not assume the software is a rounding error either way: on the cheapest row here it takes most of a year of chat before the tokens overtake the licence, while on the dearest row they pass it inside the second month. "Is Pro worth it" and "which API should I use" are separate decisions, and which one is the bigger lever depends on which row you land in. The title column is small in absolute terms, but it is the one line here you can move without changing anything you actually notice — Settings > Extensions > Summarize accepts a cheaper model.

Scale by your own habits before treating this as a budget: a user who keeps one very long thread open all week pays far more per turn than these averages, because every turn resends the history. Kunavo's catalog amount is a billing floor rather than a cap — when the upstream reports its charge, the bill is the greater of catalog cost and upstream cost times the applicable markup. Cache charges and external tools sit outside this example. The minimum top-up is $10 in prepaid credit, which funds a balance rather than buying a plan. See billing details.

Best API for ChatWise: which route wins when

Two boundaries are specific enough to change a decision. First, ChatWise's observability doc says to make sure the API base URL for the respective provider is empty, "otherwise observability will not be enabled" — so filling that field to route a provider elsewhere and keeping that particular dashboard are mutually exclusive. Second, its Azure OpenAI doc accepts only a fixed, now-dated list of deployment names, which is a concrete reason a gateway is easier here than a cloud endpoint. Both checked September 19, 2026. For the wider comparison see best LLM gateway and OpenRouter alternatives.

Pointing ChatWise at a custom endpoint

ChatWise's custom provider doc gives the path verbatim: Settings → Providers → "+" → choose a preset or create a custom configuration → configure API key and endpoint settings → add models manually or click fetch models. It states that you can integrate any OpenAI-compatible or Anthropic-compatible AI service, and that if the API supports a /models endpoint you can fetch the available models automatically instead of typing them in. Provider configuration is not one of the features the pricing page lists under Pro, and ChatWise's pricing FAQ says bringing your own key needs no subscription — but no ChatWise page states outright that the free tier can add a custom provider, so confirm that during the 7-day trial.

Read this table as Kunavo's instruction, not as ChatWise's documentation. The custom-provider doc says only "endpoint settings", no ChatWise page states whether the base URL should end in /v1, and third-party gateway guides contradict each other on it. If the fetch button returns nothing or calls come back 404, try the other form. Kunavo serves /v1/chat/completions, /v1/responses, /v1/messages and /v1/models — route files confirmed in the repository on September 19, 2026, covering the chat, Responses and Messages protocols plus the model list the fetch button needs. Whether ChatWise's fetch parser and its Responses mode agree with our implementations has not been runtime tested. This is a published configuration reference, not a compatibility test.

Set it up with a working fallback still configured, run one bounded conversation, then read the charge your account recorded for it. Kunavo has no ChatWise setup page yet; the closest published reference is the Cherry Studio integration, which documents the same OpenAI-compatible base URL for a desktop chat client, and Chatbox covers the same pattern. Create a Kunavo account when you are ready to fund a key. Comparing clients rather than providers? ChatWise vs Chatbox sets the one-time licence against a client that sells hosted model access, and AI cost optimization covers the measurement method.

FAQ

Read on September 19, 2026: ChatWise's homepage and pricing page in a browser with the FAQ answers expanded, plus its custom-provider, migration, web-search, summarize, auxiliary-visual, auxiliary-think, documents, agent, tools, GitHub Copilot, Azure OpenAI and observability docs; Anthropic's Claude Code legal page; Anthropic's and GitHub's plan pricing; the egoist/chatwise-releases GitHub API; and a direct HTTP check on the 26.9.0 Apple Silicon release asset. Not confirmed here: the changelog entries, the full provider preset list inside the app, and the exact endpoint field label in the running build. ChatWise was not installed or pointed at Kunavo, so nothing here is a compatibility test. Kunavo token rates come from the live catalog, and every dollar example on this page is illustrative token arithmetic rather than a measured bill.