vLLM Endpoint

Serve a Gemma model

Deploy a Gemma model as an OpenAI-compatible vLLM endpoint — a lightweight open model good for cost-sensitive chat/completions traffic.

Estimated $0.30–$0.55/hr while serving · Long-running endpoint — starts in 2–5 min

What inputs you need

  • Model: Gemma variant (e.g. Gemma 2 9B)
  • Max-cost budget
  • Optional: concurrency overrides

What Badgr returns

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

Recommended GPU routes

Estimated cost: $0.30–$0.55/hr while serving · Estimated runtime: Long-running endpoint — starts in 2–5 min

Example command

badgr serve google/gemma-2-9b-it \
  --gpu L40S \
  --max-cost 8

Common setup failures Badgr avoids

Assuming Gemma's smaller size means unlimited concurrency — throughput still degrades under heavy load; monitor first-token latency and scale routes accordingly.

Context length errors on long prompts — confirm Gemma's max context before routing long-context traffic to it.

Ready to run this?

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