# AI use cases for Poly Strata (VecSzn/PolyStrata) Repo: [VecSzn/PolyStrata](https://github.com/VecSzn/PolyStrata) | Use cases: 4 | Page: https://dedupedai.com/use-cases/vecszn-polystrata/ PolyStrata runs large AI models on your own Windows or Linux PC and serves them through a localhost OpenAI or Anthropic compatible API, so nothing leaves the machine. This file holds every use case of this repo. Each one starts with a "## " heading and ends with its URL. - Run Claude Code on a local model with no cloud calls: https://dedupedai.com/use-cases/vecszn-polystrata/run-claude-code-on-a-local-model/ - Point existing AI scripts at a private local API: https://dedupedai.com/use-cases/vecszn-polystrata/keep-app-and-script-ai-calls-on-your-pc/ - Ask questions about pictures without uploading them: https://dedupedai.com/use-cases/vecszn-polystrata/read-screenshots-and-photos-privately/ - Compare local models on your card before you commit: https://dedupedai.com/use-cases/vecszn-polystrata/test-which-local-model-fits-your-card/ ## How to run Claude Code on a local model with no cloud calls with Poly Strata Repo: [VecSzn/PolyStrata](https://github.com/VecSzn/PolyStrata) | Difficulty: Intermediate | Time: 1 hour | Basis: From the README Point Claude Code at Qwen3.6-35B-A3B on your own PC to edit and test code privately. **Who it is for:** A developer who wants a coding agent for a private codebase and has an NVIDIA laptop or desktop with 8 GB of VRAM or more. **The problem:** Cloud coding agents send your source code off the machine, and you may not want that or be able to pay per token for long sessions. **The outcome:** Claude Code runs against a local Qwen3.6-35B-A3B server, adds a feature, writes a test, and runs the tests on your PC. ### Steps 1. Install Poly Strata with the one-click installer for your operating system from the README. 2. Download the Qwen3.6-35B-A3B file size that fits your card, such as UD-Q3_K_M for an 8 GB laptop card. 3. Start the local server so it exposes the Anthropic API on localhost. 4. Set Claude Code to use the localhost address as its API endpoint, then give it a small coding task in your project. 5. Read the diff Claude Code produces and confirm the tests it wrote pass when it runs them. ### Prompt to paste into your agent ```text Use the PolyStrata repo at with Qwen3.6-35B-A3B served on localhost. Ask Claude Code to add a --json option to , write a test for it, and run the tests. Expected output: the code change, the new test, and the test run result. Verify that the test run passes and that no request went to a remote host. ``` ### Caveat Only Qwen3.8-Flash-Next runs on the unchanged Strata engine; the Qwen3.6-35B-A3B model runs on the experimental Poly Strata engine. 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/vecszn-polystrata/run-claude-code-on-a-local-model/ ## How to point existing AI scripts at a private local API with Poly Strata Repo: [VecSzn/PolyStrata](https://github.com/VecSzn/PolyStrata) | Difficulty: Intermediate | Time: 1 hour | Basis: From the README Redirect scripts written for the OpenAI or Anthropic API to a model on localhost. **Who it is for:** A developer or analyst with scripts that call a hosted AI API and who needs to stop sending sensitive text to a third party. **The problem:** Scripts that summarize client notes or internal documents send that text to a hosted provider on every run. **The outcome:** The same scripts keep working, but requests go to a model served on localhost and the text stays on the PC. ### Steps 1. Install Poly Strata with the one-click installer for your operating system. 2. Choose one installed model, such as Qwen3.8-Flash-Next if you have a 12 GB card, and start the server. 3. In your script, change only the base URL to the localhost address and keep the request format the same. 4. Run the script on one test document and compare the output with what the hosted API returned before. ### Prompt to paste into your agent ```text Take the script at that calls an OpenAI or Anthropic style API and point its base URL at the PolyStrata server on localhost, using model . Expected output: the script's summary of produced without any remote request. Verify that the output is a usable summary and that the network traffic goes only to localhost. ``` ### Caveat Output quality and speed depend on the model and card; the README reports speeds measured on specific PCs, not guarantees for yours. 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/vecszn-polystrata/keep-app-and-script-ai-calls-on-your-pc/ ## How to ask questions about pictures without uploading them with Poly Strata Repo: [VecSzn/PolyStrata](https://github.com/VecSzn/PolyStrata) | Difficulty: Intermediate | Time: 1 hour | Basis: From the README Send screenshots or photos to a local vision model that reads images on your PC. **Who it is for:** A user with private images, such as scanned receipts, product photos, or screenshots of internal dashboards, and a PC with a suitable card. **The problem:** Picture-reading AI services require uploading the images, which may include personal or confidential content. **The outcome:** You get answers about the pictures from a model running on your PC, and the images are never uploaded. ### Steps 1. Install Poly Strata with the one-click installer for your operating system. 2. Pick a model that reads pictures, such as Qwen3.8-Flash-Next, Gemma 4 26B-A4B, or MiniCPM-V 4.7 35B-A3B. 3. Start the server with that model and send it one image with a question through the localhost API. 4. Check the answer against the picture yourself before you rely on it. ### Prompt to paste into your agent ```text Using the PolyStrata server on localhost with , send the image and ask: . Expected output: a text answer describing the image. Verify that the answer matches what is visible in the image and that the image was not sent to any remote service. ``` ### Caveat Only Qwen3.8-Flash-Next uses the unchanged Strata engine; the picture-reading models on the other engine are experimental. 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/vecszn-polystrata/read-screenshots-and-photos-privately/ ## How to compare local models on your card before you commit with Poly Strata Repo: [VecSzn/PolyStrata](https://github.com/VecSzn/PolyStrata) | Difficulty: Intermediate | Time: 1 hour | Basis: From the README Switch between installed models on a running server to see which speed and answers suit you. **Who it is for:** A user who has several models downloaded and wants to choose one for daily work based on speed and answer quality on their own hardware. **The problem:** Published speed numbers come from other PCs, so it is hard to know which model file runs well on your card. **The outcome:** You have a short list of models and file sizes that work on your PC, with your own rough speed notes. ### Steps 1. Install Poly Strata with the one-click installer and download two or three models in different file sizes. 2. Start the server with the first model and send it the same prompt you use day to day. 3. Change to the next installed model while the server keeps running, using the model switch described in the docs. 4. Note the answer speed and quality for each model and keep the one that fits your card. ### Prompt to paste into your agent ```text With the PolyStrata server running on localhost, send the same prompt to each installed model in turn, switching models between runs. Expected output: a table of model name, approximate tokens per second, and a short quality note for each. Verify that every model answered the same prompt and that the speeds were measured on this PC. ``` ### Caveat Model switching is described in docs/BACKENDS.md, and the eight non-Qwen3.8 models are experimental on the Poly Strata engine. 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/vecszn-polystrata/test-which-local-model-fits-your-card/