ChatGPT Work users report a 15GB remote VM that can run Gemma 3 4B
rez0__ · x · 2026-07-23
A user reports that ChatGPT Work appears to provide a surprisingly capable remote VM, and that it can be abused to run local LLMs inside the environment.
- The claimed VM has about 15 GB RAM and 9 CPU cores.
- It includes a browser and can install packages, making it useful for agentic workflows without a separate VPS.
- In the demo, the model set up a proxy through ChatGPT Sites, downloaded Gemma 3 4B, installed Ollama, and ran inference in about 14 minutes.
- The benchmark screenshot shows 8.46 tokens/s generation and a 20.9s cold request.
The main takeaway is that paid ChatGPT Work users may have access to a remote machine that is much more capable than many people realize.
Related event: ChatGPT Work Offers 15GB Remote VM to Paid Users(3 posts)→
More from coding & agent
- Browserbase open-sources BELL, a full OSS stack for personal AI assistants with $10K of agentic spending — julianweisser · 2026-10-27
- Engineer's real numbers: 20B tokens in 4 months, agents still beat hiring by a mile — thechrisperry · 2026-10-27
- Vovy launches a cursor-following voice agent for Mac, free tier gives 100 credits/mo — Scobleizer · 2026-10-27
- Redditor seeks real-world stacks for a persistent-memory personal AI that answers texts and emails — BrooksomidusGG · 2026-10-27
- Practical tip: use max effort for planning, xhigh for implementation in Claude Code — majidmanzarpour · 2026-10-27
- Chrome team explains WebMCP: agents get tools from your website, not a replacement for MCP — rseroter · 2026-10-27