Cost Analysis: 8x MI325X Self-hosted Uncensored GLM-5.2
petrusenko_max · x · 2026-07-07
petrusenkomax estimated the operational plan for deploying a private, uncensored GLM-5.2 (approx. 128K context) using 4-bit quantization on an 8×MI325X (256GB) node: targeting 100 subscribers at about $200 per month.
Full load estimates: 20 tokens/s per user with 10 active users; dropping to 10 tokens/s with 50 users.
Given sufficient demand, this configuration can support exclusive access to large-context foundation models, making it suitable for scenarios requiring private, uncensored, high-context LLM services.
More from Infra
- CoreWeave 5-year CDS spikes to ~1,100bps as market fears AI bubble starts as credit event — kimmonismus · 2026-10-26
- NVIDIA Plans ~$2,000 Per GPU Per Year Software Fee for Running AI Agents — LinusEkenstam · 2026-10-26
- 12 DevOps concepts to learn in order, from Linux to SRE — _jaydeepkarale · 2026-10-26
- Killed our Vector DB for SQLite BM25 and wrote a 3.6ns Rust sentinel: lessons from multi-agent loops — OpenComplaint524 · 2026-10-26
- Gemma 4 26B-A4B runs offline on a 16 GB Mac: 33/40 on TypeScript tasks at 31 tok/s — Mission-Sir-1614 · 2026-10-26
- How thousands of GPUs train one AI: a breakdown of 5D parallelism — jbhuang0604 · 2026-10-26