> Markdown twin of https://conduitllm.com/providers/openrouter.html > Use OpenRouter in a free, open-source desktop app. Setup in four steps, what it costs, free-model limits, web search, and what doesn't work yet. --- Provider guide # Use OpenRouter in a desktop app OpenRouter gives you one API key for hundreds of models from many companies — Claude, GPT, Gemini, DeepSeek, Kimi, GLM, Qwen, MiniMax and more — paid from one balance. Conduit is a free, open-source desktop app for that key: switch models mid-conversation, use MCP tools, ask your own documents, and keep every conversation on your own disk. OpenRouter details below were checked against OpenRouter's own documentation on 4 October 2026; if something has changed since, their documentation is right. Part of the [provider guides](https://conduitllm.com/providers.html). ## Set it up 1. **Get a key.** Sign in at OpenRouter, create a key on its [API keys page](https://openrouter.ai/settings/keys), and add credits if you want paid models. Conduit has no link to this page yet, so open it in your browser. 2. **Pick OpenRouter.** In Conduit, open **Settings → Providers & keys** and choose **OpenRouter** from the **Provider** list. If local-only mode was on, Conduit turns it off and tells you, because OpenRouter needs the internet. 3. **Save the key.** Paste it into **Provider secret** and choose **Save provider key**. It goes into your operating system's keychain, and the interface never sees it again. 4. **Pick a model.** Choose **Load models**, then pick from the **Model** list. Conduit shows every model OpenRouter offers your key, so the list is long; to search it, open the command palette with Ctrl+K and type / followed by part of the model's name. From then on, the model menu under the message box lists your OpenRouter models next to any other provider you have set up, and Ctrl+Shift+P (⌘+Shift+P on macOS) switches provider without starting a new chat. ## What works through OpenRouter | Feature | With OpenRouter | | --- | --- | | Chat | Yes, streamed. The model's reasoning shows when it sends any. | | MCP tools and built-in tools | Yes, on models that support tool calling. A tool asks before it runs unless it is marked read-only. | | Images in your messages | Sent to every OpenRouter model; works on models that accept images. | | Web search | Yes, through OpenRouter's own search, billed to your key, with sources shown. Off by default; asks before the first search. | | Image generation | Yes, billed to your key. Conduit asks once before the first image and saves each image on your disk. | | Your documents | Yes. Each document's text is sent once to OpenRouter to be indexed, after you agree; answers cite the passages they used. | | Personal apps, Slides, workflows | Yes, like any other provider. | | Token counts | Yes, per reply, including cached tokens. | | Cost in dollars | Not yet. See "What it costs" below. | ## What it costs Conduit is free and adds nothing. OpenRouter says it passes through each model's price with "no markup on inference pricing," and charges a fee when you buy credits: 5.5% with a $0.80 minimum by card, or 5% in crypto. Prices differ by model and are listed per million tokens on [OpenRouter's models page](https://openrouter.ai/models). - **Free models** (ids ending in `:free`) are limited to 20 requests a minute and 50 a day, rising to 1,000 a day once you have bought at least 10 credits. - **Web search costs extra, even on free models.** For models searched through Exa, OpenRouter charges $0.007 per search; OpenAI, Anthropic and Google models use their own search at their own prices. - **Seeing your spend.** Conduit shows the tokens each reply used, but this release does not turn OpenRouter usage into a dollar figure. Your exact spend is on OpenRouter's [activity page](https://openrouter.ai/activity). Whether paying per token beats a subscription depends on how much you use it; the [API vs subscription calculator](https://conduitllm.com/api-vs-subscription.html) works it out from your own numbers. ## Where your data goes Your conversations are stored on your own disk. When you send a message, Conduit sends it from your machine to OpenRouter, which passes it to the company serving that model. OpenRouter states that prompts and completions "are not logged by default" unless you opt in; the serving company's own policy applies after that. Conduit has no telemetry and no Conduit server in between. Turn on local-only mode and Conduit refuses OpenRouter and every other cloud provider. ## What doesn't work yet - **No dollar costs for OpenRouter models** — token counts only; check OpenRouter's activity page for spend. - **No capability labels** — the model list doesn't mark which models handle tools or images, so a model that doesn't will ignore them or return an error. - **No custom endpoint** — OpenRouter's address is fixed; for a proxy in front of it, use the OpenAI Compatible provider. - **No in-app key link** — create the key on OpenRouter's website. - **Image model is fixed** — image generation uses Conduit's default OpenRouter image model, not one you choose. [Download Conduit](https://github.com/runningpixels/conduit/releases) Other providers: [all provider guides](https://conduitllm.com/providers.html). Why people bring their own key: [pay per token, not per seat](https://conduitllm.com/use-cases/bring-your-own-key.html). Connecting tools: [Conduit as your MCP client](https://conduitllm.com/use-cases/mcp-client.html). ## Questions **Does Conduit charge anything on top of OpenRouter?** No. Conduit is free and open source, with no account and no subscription. Your requests go from your machine to OpenRouter, and OpenRouter bills your OpenRouter credits. Conduit never sees the bill and takes nothing from it. **Can I use OpenRouter's free models in Conduit?** Yes. Free models appear in the model list like any other, and their ids end in `:free`. OpenRouter limits them to 20 requests a minute and 50 a day, or 1,000 a day once you have bought at least 10 credits. Web search still costs money on a free model. **Does web search work with OpenRouter?** Yes, once you switch web search on. Conduit uses OpenRouter's own web search, billed to the same key, and shows the sources the answer used. OpenRouter runs the model maker's search for OpenAI, Anthropic and Google models and Exa for the rest. Web search is off by default and asks before its first use. **Does OpenRouter keep my prompts?** OpenRouter says it does not log prompts or completions unless you opt in. Your request is then passed to the company that serves the model, under that company's own policy. Conduit itself sends nothing anywhere else and has no telemetry. **Do MCP tools work with OpenRouter models?** Yes, on models that support tool calling. Conduit sends your tools to whichever model you pick; a model without tool support will ignore them or return an error. Conduit lists every model OpenRouter offers and does not yet mark which ones support tools. **Can I use MiniMax or other models Conduit has no provider for?** Yes. Any model OpenRouter offers your key appears in Conduit's model list, including MiniMax, which has no provider of its own in Conduit. Sources for the OpenRouter facts, checked 4 October 2026: OpenRouter's [FAQ](https://openrouter.ai/docs/faq) (fees, pricing pass-through, free-model limits, logging), [rate limits](https://openrouter.ai/docs/api-reference/limits), [web search](https://openrouter.ai/docs/features/web-search), and [web search pricing](https://openrouter.ai/docs/guides/features/server-tools/web-search). Conduit rows match v1.0.0-rc.7 and are checkable in [the documentation](https://conduitllm.com/docs.html) and [the source](https://github.com/runningpixels/conduit). Found an error? [Tell us](https://github.com/runningpixels/conduit/discussions) and we will fix it. ## Related - [Use the DeepSeek API in a desktop app](https://conduitllm.com/providers/deepseek.html) — Use the DeepSeek API in a free, open-source desktop app. Setup, peak and off-peak prices, thinking mode, tools, where your data goes, and current limits.