{
 "schema_version": 1,
 "generated": "2026-10-31",
 "id": "startluxlabs-startlux-decision--run-27b-autothink-on-one-gpu-with-prism",
 "repo": "StartLuxLabs/StartLux-Decision",
 "repo_url": "https://github.com/StartLuxLabs/StartLux-Decision",
 "repo_slug": "startluxlabs-startlux-decision",
 "title": "Run the 27B AutoThink decision model on one GPU from a GGUF",
 "page_title": "How to run the 27B AutoThink decision model on one GPU from a GGUF with Start Lux Decision",
 "summary": "Use the StartLux-Prism server to serve the 27B AutoThink Q8 GGUF on a single CUDA GPU.",
 "who": "A developer with one large GPU who wants the most accurate StartLux model without a multi-GPU vLLM setup.",
 "problem": "The full 27B model with vLLM expects several GPUs, which most teams do not have.",
 "outcome": "A local /v1/systemone server on port 8090 answering decisions with the 27B AutoThink model.",
 "steps": [
  "Run `git clone https://github.com/StartLuxLabs/StartLux-Prism.git` and `cd StartLux-Prism`",
  "Create the environment with `python3.12 -m venv .venv && source .venv/bin/activate` and install requirements.txt",
  "Download the Q8_0 GGUF and the model config files with the two hf download commands from the README",
  "Start the server: `python -m startlux_prism.server --gguf $M/StartLux-Decision-v1.1-27B-AutoThink-Q8_0.gguf --model-dir $M --port 8090`",
  "Send a test request to localhost:8090/v1/systemone"
 ],
 "agent_prompt": "Set up StartLux-Prism for StartLuxLabs/StartLux-Decision on this Linux GPU box, download the 27B AutoThink Q8_0 GGUF, start the server on port 8090 and send the README's ticket example to confirm it answers.",
 "agents": [
  "claude-code",
  "codex"
 ],
 "difficulty": "Advanced",
 "time": "1 hour",
 "basis": "readme",
 "caveat": "Prism needs Linux, Python 3.12 and a CUDA 13 driver, and v1.1 Instant models are not supported by it yet.",
 "category": "Work and operations",
 "tags": [
  "Work and operations"
 ],
 "lang": "python",
 "requirements": [
  "Python environment that can install the package with vLLM support, or a Bamboo API key for the hosted model",
  "A GPU for the local vLLM setup; the README numbers were measured on one H200",
  "Disk space and network access to download a model from Hugging Face"
 ],
 "risk": "none",
 "seen_in_the_wild": false,
 "mentioned_publicly": false,
 "risks": [],
 "best": false,
 "score": 0.75,
 "evidence": [],
 "url": "https://dedupedai.com/use-cases/startluxlabs-startlux-decision/run-27b-autothink-on-one-gpu-with-prism/",
 "md": "https://dedupedai.com/use-cases/startluxlabs-startlux-decision/run-27b-autothink-on-one-gpu-with-prism/index.md",
 "disclaimer": "Deduped lists open-source repositories that belong to other people. We do not own, maintain, audit, or vouch for any of them, and a repo appearing here is not an endorsement or a safety check. The use cases, steps, and prompts are written by AI and may be incomplete, outdated, or just wrong. Where a use case is marked \"Seen in the wild\", we link to public posts where someone described something similar, but we have not verified those posts and the use case itself is still our AI's own write-up. Use all of this at your own risk. Only run code, commands, or agents on systems and accounts you are authorized to use, and read the code and every command before you run it, especially anything that touches credentials, files, or the network. Nothing here is professional, security, legal, or financial advice. All trademarks and project names belong to their respective owners, and their use here does not imply affiliation. Content is provided as is, without warranties of any kind. Found a mistake, a safety problem, or something you want removed? Email hello@dedupedai.com and we will look at it promptly.",
 "api": "https://dedupedai.com/api/v1/usecases/startluxlabs-startlux-decision--run-27b-autothink-on-one-gpu-with-prism.json",
 "api_version": "v1"
}