Anthropic's desktop app for macOS, Windows and Linux (beta), for chat, with Claude Code and Cowork inside it on paid plans and with desktop extensions for local files and tools.
Our description, from the vendor's documentation, as of 2026-10-22; reasoned from those pages, not observed running.
Runs onThe places the vendor's own pages say the product runs: your machine, a container environment you run, or a cloud service the vendor runs. It is not a limit.
Models runWhere the vendor's own pages say the product's models run, not a limit: your machine, a container environment you run, the model company's cloud (servers of the company that makes or serves the model), or a cloud provider's model service. For a cloud, read the data and compliance lines of the company running it, linked by its name.
Several endpoints built inThe tool itself can hold more than one provider or address, so nothing has to sit in front of it to reach different providers.
Same role (chat front end) at the same layer (Coding agents and assistants). Up to 4 others, in alphabetical order; nothing is ranked.
Product
Several providersOne session can use models from different providers, for example one agent on one maker's model and another on a second maker's. Switching the single model for the whole session does not count on its own. A tool that is itself a gateway or library routing to the providers its page names counts.
Several endpointsThe tool itself can hold more than one provider or address, so nothing has to sit in front of it to reach different providers.
Model on your machineThe tool can use a model running on your own computer, for example through Ollama, LM Studio, oMLX or llama.cpp.
Local and cloudA local model (on your machine or in a container environment you run) and a cloud model (in its model company's cloud or a cloud provider's model service) can both be used in the same session or setup.
Words in the cells: YesA page we read says so., PartlyA page says part of it, for example separate models per agent without saying the providers can differ., Via a proxyOnly with a proxy, or a local server that imitates a cloud provider's interface, in front of the tool., NoA page we read says the product does not support it., ClaimedA vendor page says so on a marketing or overview page, and we found no guide showing how to set it up., Not foundWe read pages for it and found nothing there. It says nothing about whether the product has the feature., n/aThe question does not apply to the product, for example because it is a model server and not a tool that connects to models., BlankNothing is published for it yet.
Data and compliance
Compliance: Not checked yetWe have not read a page of the right kind for this question yet.: no page that could answer it is recorded as read.
Data: Training use, default retention, zero-retention option and data residency not checked yetWe have not read a page of the right kind for this question yet..
Model and where it runs: Installed: a desktop application; model provider and where it can run not checked yetWe have not read a page of the right kind for this question yet..
Details: What the maker's own pages state, not a compliance determination.
What the maker's own pages said on the dates shown, not a compliance determination.
claude.com page: download(the product's own site, from our list of products; not read for these lines)
The readings behind this page
Our summary of the vendor's pages as of the dates shown on each line (2026-10-19 to 2026-10-24); confirm on the vendor's page.
What you set up
Ollama's blog post on Claude Desktop says that turning on the Claude option in Ollama's settings makes Ollama set up the connection for Claude Desktop, with Ollama acting as the third-party gateway. (ollama.com page, read 2026-10-20)
Anthropic's configuration page says that when managed (MDM) configuration is present for Claude Desktop's third-party mode, values written locally are ignored, unless the managed source sets only app-behavior keys; then all other settings remain local and users can edit them. (claude.com page, read 2026-10-19)
One sentence naming both a local and a cloud path
Ollama's blog post on Claude Desktop says that while Claude Desktop is connected to Ollama, any model within Ollama can be chosen, whether run on the user's computer or on Ollama's cloud. (ollama.com page, read 2026-10-20)
Account or plan
Switching between Ollama and Anthropic models is possible by using the toggle in Ollama. (ollama.com page, read 2026-10-20)
Models from different providers in one session
On Anthropic's gateway page for Claude Desktop's third-party mode, when no model list is set the app fills its model picker from the gateway and keeps only models whose IDs look like Claude's; setting a model list replaces that discovery. (claude.com page, read 2026-10-19)
Several endpoints, built in (no proxy)
Anthropic's gateway page for Claude Desktop's third-party mode says a self-hosted LLM gateway, such as LiteLLM, Portkey or a company's own proxy, can be the inference provider when inferenceProvider is set to gateway and a base URL and credentials are supplied. The gateway must implement Anthropic's Messages API (streaming and tool use required), and the page describes it as serving Claude models. (claude.com page, read 2026-10-19)
Claude Desktop's inferenceProvider setting picks one inference backend (gateway, Anthropic, Bedrock, Mantle, Vertex or Foundry), and setting it switches the app into third-party mode. (claude.com page, read 2026-10-19)
Claude Desktop's first-party API page says that setting inferenceProvider to anthropic sends inference traffic to Anthropic's API endpoints instead of keeping it inside a cloud provider. (claude.com page, read 2026-10-24)
Claude Desktop's third-party overview describes a deployment mode in which all model inference goes through a provider the user configures, which can be the Anthropic API directly, a compatible gateway the user operates, Microsoft Foundry, Amazon Bedrock or Google Cloud's Agent Platform. (claude.com page, read 2026-10-24)
What a local model must meet
Anthropic's gateway page says a gateway in front of Claude Desktop's Cowork and Code sessions should keep prompt caching: sessions send cache_control markers each turn, and a gateway that drops them, or changes the opening instructions or the list of tools between requests, makes the upstream provider reprocess the whole conversation every turn, with more input tokens and more delay. (claude.com page, read 2026-10-19)
Anthropic's gateway page for Claude Desktop on third-party inference says the gateway must implement the Anthropic Messages API, and lists the messages endpoint with streaming and tool use as required. (claude.com page, read 2026-10-19)
Anthropic's gateway page says that in Claude Desktop on third-party inference a session assumes a 200K-token window for any model ID the app cannot match to a native 1M-token window. (claude.com page, read 2026-10-19)
Plan, policy and billing
Anthropic's download page answers 'no' to needing a paid plan for its desktop or mobile apps: they can be used on Free, Pro, Max, Team or Enterprise, though some features need a paid plan. (claude.com page, read 2026-10-22)
For the Linux build, Anthropic's download page says Chat is on every plan, Free included, while Claude Code and Claude Cowork need Pro, Max, Team or Enterprise. (claude.com page, read 2026-10-22)
Anthropic's overview page for Claude Desktop on third-party says the provider you pick takes care of inference and billing. (claude.com page, read 2026-10-23)
Anthropic's gateway page describes two ways to authenticate to the gateway: a shared gateway API key placed in the managed configuration, or per-user sign-in through the organization's identity provider, after which the app sends a per-user token to the gateway. (claude.com page, read 2026-10-19)
A separate model for planning and for doing
The app uses tier mapping (anthropicFamilyTier) to identify the Claude tier an entry represents, allowing bare tier aliases to resolve to the customer's model. (claude.com page, read 2026-10-19)
About the product
Anthropic's help article describes the Claude desktop apps as bringing Claude to your computer. (support.claude.com page, read 2026-10-22)
macOS, Windows, and Linux (beta) are the platforms on which Claude Desktop is available. (support.claude.com page, read 2026-10-22)
Claude Code executes within the desktop application. (claude.com page, read 2026-10-22)
The desktop app stays accessible from your dock, offers quick entry for instant system-wide access, and provides desktop extensions to link Claude with local tools and files. (claude.com page, read 2026-10-22)
Standard enterprise deployment is supported by the Claude desktop app using MSIX installers for Windows and PKG installers for Mac. (claude.com page, read 2026-10-22)
Our reading of vendors' pages, as of the dates shown. Confirm on the vendor's page before relying on any line that cites it. To report an error write to report@crinaro.ai