llm.bunkerchain.io - Bunkerchain office DGX Spark, OpenAI-compatible

Put the model in the path and use it as your base_url:

  https://llm.bunkerchain.io/qwen3.8-27b/v1      27B, vision-capable, ~15 tok/s
  https://llm.bunkerchain.io/qwen3-30b-a3b/v1    30B MoE, text only, ~50 tok/s

  curl https://llm.bunkerchain.io/qwen3.8-27b/v1/models

The model runs on-premises in the Bunkerchain office; this host only proxies.