Connect your own model API keys to an organized chat workspace with agents, reusable prompts, project context and parallel responses. Personal licenses are one-time purchases; model usage and optional cloud services have separate costs.
Used this tool? Rate it
Used this tool? Rate it
TypingMind is an AI chat interface that connects to large language models through your own API keys, bringing ChatGPT, Claude, Gemini and other models into one workspace.
A personal license unlocks software features; it does not supply free model tokens. AI providers bill API usage separately, and optional cloud storage is another service.
AI agents are configurable extensions of a model's initial system instructions. They can specify a base model, generation parameters, plugins and a text-to-speech voice.
Agent knowledge has three documented routes: directly attached training files, Dynamic Context that retrieves content from an API and injects it into the system prompt, and a knowledge base that retrieves relevant uploaded material. Directly attached agent documents enter the system instructions and must fit the chosen model's context length.
Tools include provider-native web search and a provider-native Python code sandbox, alongside installable plugins. The documented Interactive Canvas plugin can produce forms, HTML/JS/CSS, games and visualizations, and can display images from a chat.
These are vendor-described workflows, not measured accuracy or productivity improvements.
Individuals who manage their own model access are the audience for the personal license. The local version needs no account login: the documented access pattern uses a license key and API key. It has no team-management panel.
Teams and businesses needing shared administration are the audience for Teams. Administrators can preset API keys, build prompts, plugins and agents, connect company knowledge, restrict member access, invite members by email and customize branding and a domain.
A bulk key increases licensed users and devices; the separate Teams product supplies administration.
The current Setapp listing labels TypingMind Mac, Web, version 1.5.6, requiring macOS 11.3 or later. The vendor's separate distribution FAQ describes supplied GPT-3.5/GPT-4 models without a user API key, optional personal keys and usage limits. It restricts supplied models and premium features to the macOS Setapp app, excluding the web app, and says no free Claude model is included yet. The catalog's Web label does not by itself extend those supplied-model terms to web access. Distributor support describes AI+ credits shared across its AI apps, varying by model and interaction; unused credits reset each billing period.
Teams has a distinct Teams Self-host product, separate from the personal Static Self-host package.
The personal price cards captured on October 29, 2026 display the following one-time amounts. The seller uses the $ symbol without an ISO currency code in the captured pricing HTML. Features build on the preceding tier.
| Personal plan | Displayed price | Listed scope |
|---|---|---|
| Standard | $39 one time | Remove ads/popups; basic chat; AI agents; chat/plugin sharing; voice input |
| Extended | $79 one time | Standard plus web search, vision/images, image generation, text-to-speech and document uploads |
| Premium | $99 one time; 50% OFF, with $198 crossed out | Extended plus multi-model chats, projects/folders, artifacts/canvas, unlimited plugins/MCPs and free updates |
| Bulk license | $395 one time; 50% off, with $790 crossed out | Shared Premium-benefit key for up to 10 users on 50 devices |
Personal terms cover one user, the hosted app and static self-host, with a license valid without renewal. The Premium FAQ specifically promises all updates forever; the EULA differs, as described under Limitations.
An individual key allows five devices. Separate browsers count separately: Chrome and Firefox on one laptop have independent histories and consume two activations. The FAQ describes unlinking devices through License Manager. Personal terms still limit the individual license to one user.
API costs and extra cloud storage are excluded. Optional TypingMind Cloud includes 50MB free, described as typically enough for 5000 chats, with 1GB and 5GB upgrades payable monthly or annually. Actual chat capacity varies with content size. The checked storage guide does not display upgrade prices.
The captured Teams page had its annual billing switch selected. It showed Starter at $83 per month with five seats and additional seats at $8 per seat per month, Growth at $166 per month, and Professional at $249 per month, with $499 crossed out and a 50% OFF, “Ends soon” offer. These are the monthly figures in the annual-selected display; a monthly-billed selection was not captured. API usage is excluded, and seat and knowledge-base capacity add-ons are separately available.
Team-license terms describe a limited license period with automatic renewal and non-refundable purchases, except mandatory consumer rights. These terms differ from the personal refund offer. The personal self-service refund page allows a request within 14 days of purchase, begins with the purchase email, and says automatic refunds return within 3–5 business days. The personal terms also preserve applicable mandatory consumer rights.
The distributor listing separately advertises membership from $14.99/month and a 7-day trial, covering TypingMind and other apps. This is a membership offer, not a personal web license or a statement that all model usage is included.
Two independently discovered alternatives illustrate different deployment choices:
These vendor descriptions establish deployment, customization and administration differences, not independent performance, security or total-cost rankings.
The privacy policy says local/static mode keeps API keys, chat history, prompts and messages in the browser's Local Storage and IndexedDB. Data is separate for each browser or device, and the company says it cannot access, view or recover that locally stored data.
By default, prompts, messages and attachments go directly from the browser to the selected AI provider. An optional built-in proxy is available when a provider does not accept direct browser requests. The policy says the proxy passes requests through without reading or logging requests or responses. Data received by AI providers remains subject to their own privacy policies.
There is an unresolved inconsistency about keys in the public documentation. The privacy policy says API keys stay local and are never transmitted to TypingMind's servers. The Cloud Sync guide, however, lists API Keys and License key among data stored on the Cloud server when synchronization is enabled, together with chats, prompts, agents, plugins, settings and attachments. It says Sync Settings can choose which categories synchronize. These statements do not establish a single local-only rule for optional sync configurations.
Opting into Cloud or using Teams involves server processing and storage. The privacy policy lists account/contact and license information, selected cloud content, IP addresses, anonymous crash logs and aggregated website analytics. It says communication uses HTTPS and AWS cloud databases are encrypted at rest. Proxy-content logging and operational/account data have different scopes.
Optional password encryption is specifically described for API keys. The pricing FAQ says chat messages, prompts and agents use standard browser local storage, which may or may not be encrypted by the browser; key encryption must also be configured on each device. This does not promise TypingMind encryption for every local conversation.
Cloud sharing stores the shared material on its servers and makes it publicly accessible through a secret link, with search indexing prevented. Deleted chats, agents, prompts, plugins and folders remain in Cloud's Recently Deleted area for 30 days before permanent deletion. Separately, the privacy policy says accounts without a paid subscription that have been inactive for 90 days will have data deleted with advance notice.
The vendor's data-use policy describes conversations on the TypingMind system as confidential and never used for training, while explicitly discussing TypingMind Custom and stating that selected providers determine their own data-use and training policies too. This is not a universal no-training promise from connected providers or their free tiers.
The terms state that users retain ownership of their uploaded and generated content while remaining responsible for it. They also disclaim the accuracy, completeness and suitability of AI output and apply third-party terms to integrations. The separate restrictions on modifying, redistributing and reselling the application still apply.
The pricing FAQ directs purchasers to the license-recovery page.
Yes. Settings → App data & storage provides export/import controls, with a choice of data categories. Individual chats can be exported as JSON through their menu, and JSON files can be imported.
No. The installation guide says the PWA shares the browser's local storage, account, settings and keys.
