vLLM Endpoint

Serve a DeepSeek model

Deploy a DeepSeek model as an OpenAI-compatible vLLM endpoint — strong reasoning and code performance for chat/completions traffic.

Run this jobGet help running this

Estimated $0.80–$2.50/hr while serving, depending on model size · Long-running endpoint — starts in 3–7 min

What inputs you need

  • Model: DeepSeek variant (e.g. DeepSeek-R1-Distill or DeepSeek-Coder)
  • Max-cost budget
  • Optional: context-length override

What Badgr returns

  • OpenAI-compatible endpoint URL + API key
  • Health-check status
  • Usage receipt on teardown

Recommended GPU routes

H100 80GBA100 80GBL40S 48GB

Estimated cost: $0.80–$2.50/hr while serving, depending on model size · Estimated runtime: Long-running endpoint — starts in 3–7 min

Example command

badgr serve deepseek-ai/DeepSeek-R1-Distill-Qwen-32B \
  --gpu H100 \
  --max-cost 15

Common setup failures Badgr avoids

Context length errors on long-reasoning prompts — confirm the model's max context and route to a higher-VRAM GPU for long-context workloads.

Endpoint fails health check on cold start — startup timeout accounts for the model's weight-load time.

Ready to run this?

Launch from the dashboard, CLI, or Compute API. Max-cost protection included.

Run this jobBack to vLLM Endpoint