# AI use cases for Slotstream (carloslfu/slotstream) Repo: [carloslfu/slotstream](https://github.com/carloslfu/slotstream) | Use cases: 4 | Page: https://dedupedai.com/use-cases/carloslfu-slotstream/ Slotstream is a native Swift command-line tool and local server that runs the 125-billion-parameter Qwen3.8-Flash-Next model on Apple Silicon Macs with 16 to 64 GB of memory by streaming experts from the SSD. This file holds every use case of this repo. Each one starts with a "## " heading and ends with its URL. - Chat with a large model offline on your Mac: https://dedupedai.com/use-cases/carloslfu-slotstream/private-offline-chat-on-mac/ - Code with Claude Code on a local model: https://dedupedai.com/use-cases/carloslfu-slotstream/run-claude-code-on-local-model/ - Use Codex with a local model on your Mac: https://dedupedai.com/use-cases/carloslfu-slotstream/run-codex-on-local-model/ - Connect your own app to a local model API: https://dedupedai.com/use-cases/carloslfu-slotstream/connect-own-app-to-local-api/ ## How to chat with a large model offline on your Mac with Slotstream Repo: [carloslfu/slotstream](https://github.com/carloslfu/slotstream) | Difficulty: Beginner | Time: 3 hours | Basis: From the README Install Slotstream and get a first offline reply from a 125B model on a 16 to 64 GB Mac. **Who it is for:** A Mac owner who wants a capable AI assistant without a cloud account and whose Mac has less memory than the model normally needs. **The problem:** Large open models usually need far more memory than a typical Mac has, and cloud chat services send prompts off the machine. **The outcome:** A downloaded model that answers questions on the Mac with no Python setup and no cloud account after the one-time download. ### Steps 1. Open Terminal and run `curl -fsSL https://raw.githubusercontent.com/carloslfu/slotstream/main/install.sh | sh` 2. If `slotstream` isn't found, open a new terminal window. 3. Run `slotstream doctor` to check memory and disk space without loading the model. 4. Run `slotstream run --prompt "Why is the sky blue?"` and let the first run download the model. 5. Ask follow-up questions with new `slotstream run` commands once the reply prints. ### Prompt to paste into your agent ```text In the carloslfu/slotstream repo, install the tool, run `slotstream doctor`, and then run `slotstream run --prompt ""`. Expected output: a text reply from the local model. Verify that the reply prints after the model download finishes and that no cloud service is contacted. ``` ### Caveat The first run downloads a model of about 105 GB, so the setup time depends mostly on your internet speed and free SSD space. Disclaimer: Third-party repos we don't own or vouch for. AI-generated use cases can be wrong: review everything and run at your own risk. Full text: https://dedupedai.com/disclaimer/ URL: https://dedupedai.com/use-cases/carloslfu-slotstream/private-offline-chat-on-mac/ ## How to code with Claude Code on a local model with Slotstream Repo: [carloslfu/slotstream](https://github.com/carloslfu/slotstream) | Difficulty: Intermediate | Time: 3 hours | Basis: From the README Start Claude Code connected to the Slotstream local model with one command. **Who it is for:** A developer who uses Claude Code and wants coding help that runs on their own Mac instead of a hosted model. **The problem:** Coding agents normally send source code to a hosted model, which may be unacceptable for private projects. **The outcome:** Claude Code running against the local Qwen3.8-Flash-Next model, with the Slotstream server started in the background. ### Steps 1. Install Claude Code separately, since Slotstream provides the model and not the agent interface. 2. Run the Slotstream install command from the install section of the README. 3. Run `slotstream doctor` to confirm your Mac meets the memory and disk requirements. 4. Run `slotstream launch claude` to start Claude Code on the local model. 5. Ask Claude Code to read a file in your project and summarize it to confirm the connection works. ### Prompt to paste into your agent ```text Using the carloslfu/slotstream repo, run `slotstream launch claude` in the project folder . Expected output: Claude Code starts connected to the local Slotstream model. Verify by asking it to summarize one file in and confirming a reply comes back without a network call to a hosted model. ``` ### Caveat The README says the local model is slower than hosted models on larger Macs with more memory, and reply speed depends on your Mac and workload. Disclaimer: Third-party repos we don't own or vouch for. AI-generated use cases can be wrong: review everything and run at your own risk. Full text: https://dedupedai.com/disclaimer/ URL: https://dedupedai.com/use-cases/carloslfu-slotstream/run-claude-code-on-local-model/ ## How to use Codex with a local model on your Mac with Slotstream Repo: [carloslfu/slotstream](https://github.com/carloslfu/slotstream) | Difficulty: Intermediate | Time: 3 hours | Basis: From the README Launch Codex against the Slotstream model so code tasks stay on your machine. **Who it is for:** A developer who uses Codex and wants code tasks handled on a private local model. **The problem:** Sending proprietary code to a remote model is not always allowed, and Codex needs a model it can reach. **The outcome:** Codex connected to the local model through one launch command, with the server started automatically if none is running. ### Steps 1. Install Codex separately, since it provides the agent interface and Slotstream provides the model. 2. Run the Slotstream install command and then `slotstream doctor`. 3. Run `slotstream launch codex` from the project folder you want to work in. (inferred, check the README) 4. Give Codex a small task in that folder, such as adding a comment to one function. 5. Check that the change appears in the file and that the reply came from the local server. ### Prompt to paste into your agent ```text In the carloslfu/slotstream repo, run `slotstream launch codex` from . Expected output: Codex opens connected to the local model. Verify by asking it for a one-line code explanation of and confirming it responds. ``` ### Caveat The README says only one generation runs at a time, so other apps connected to the same server will share the model and wait their turn. Disclaimer: Third-party repos we don't own or vouch for. AI-generated use cases can be wrong: review everything and run at your own risk. Full text: https://dedupedai.com/disclaimer/ URL: https://dedupedai.com/use-cases/carloslfu-slotstream/run-codex-on-local-model/ ## How to connect your own app to a local model API with Slotstream Repo: [carloslfu/slotstream](https://github.com/carloslfu/slotstream) | Difficulty: Advanced | Time: 3 hours | Basis: From the README Run the Slotstream server and point an Ollama or OpenAI client at it. **Who it is for:** A developer building an app or script that calls an LLM and wants a local endpoint with no external API key. **The problem:** Prototyping an AI feature usually means paying for a hosted API and sending test data off the machine. **The outcome:** A local server on the Mac that answers requests from Ollama-style or OpenAI-style clients. ### Steps 1. Run the Slotstream install command and then `slotstream doctor`. 2. Run `slotstream serve --prefix-cache-dir ` with a folder you choose for saved conversation prefixes. (inferred, check the README) 3. Point your Ollama or OpenAI client at the local server address shown in the startup output. 4. Send one test request from your app and confirm a reply comes back. 5. Read the engineering guide for the API references and command options before going further. ### Prompt to paste into your agent ```text In the carloslfu/slotstream repo, run `slotstream serve --prefix-cache-dir ` and send one OpenAI-compatible chat request to the local address printed at startup. Expected output: a JSON chat response from the local model. Verify that the response contains generated text and that the server log shows the request. ``` ### Caveat The README notes that the server handles one generation at a time, so load tests with many simultaneous requests will queue. Disclaimer: Third-party repos we don't own or vouch for. AI-generated use cases can be wrong: review everything and run at your own risk. Full text: https://dedupedai.com/disclaimer/ URL: https://dedupedai.com/use-cases/carloslfu-slotstream/connect-own-app-to-local-api/