Rent L40S 48GB GPU capacity

Run models, scripts, containers, training jobs, batch inference, and custom GPU workloads on L40S 48GB-class capacity with max-cost protection.

Run on L40SServe modelGet help running this

Checking last-seen pricing...

Available regions

United StatesEuropeAsia Pacific

Popular L40S 48GB use cases

image generationvideo processingComfyUIFluxSDXL

Image and video generation, vision AI workloads

Best workloads for L40S 48GB

ComfyUI

Run ComfyUI image and video workflows — SDXL, Flux, ControlNet, and custom node graphs — on GPU capacity with max-cost protection.

vLLM Endpoint

Serve an OpenAI-compatible vLLM endpoint for open models with health checks, live logs, and automatic teardown on idle or budget limit.

Batch Inference

Run offline batch inference — scoring, evals, extraction — across large input sets with retries and a single receipt at completion.

Models people serve on L40S 48GB

Qwen 2.5 7B InstructQwen 2.5 Coder 7BLlama 3.1 8B InstructMistral 7B Instruct

Example commands

Serve a model

badgr serve Qwen/Qwen3-32B \
  --gpu L40S \
  --max-cost 10

Starts an OpenAI-compatible endpoint with max-cost protection

Run a custom job

badgr run "python job.py" \
  --gpu L40S \
  --max-cost 10 \
  --max-runtime 120

Runs a script with cost and time limits, logs streamed back

Estimated pricing & availability

Checking last-seen pricing...

Prices below are last-seen estimates. Badgr rechecks live price and availability before launch — final cost depends on selected route, runtime, region, and workload duration.

Loading routes...

Frequently asked questions

Can I rent L40S 48GB for non-AI workloads?

Yes. Badgr can run AI workloads, CUDA jobs, scripts, containers, batch jobs, and other GPU workloads. L40S 48GB is popular for AI because that's the most common use case, but you can run any CUDA-compatible workload.

Can I run custom Docker containers on L40S 48GB?

Yes. Use badgr run with a custom Docker image, or submit a container through the Compute API. Badgr handles orchestration, networking, logs, and teardown.

Does Badgr enforce max-cost limits?

Yes. Set a max-cost budget before launch. If the job would exceed it, Badgr stops the job and issues a receipt. This protects you from runaway costs.

Is L40S 48GB availability guaranteed?

No. Badgr rechecks availability immediately before launch. If the route is no longer available, we notify you and suggest alternatives. This is why we show live pricing.

Do I get logs and receipts?

Yes. Stream logs in real-time while the job runs. After completion, Badgr issues a detailed receipt with compute time, cost, and resource usage.

Can I serve models on L40S 48GB?

Yes. Use badgr serve to launch an OpenAI-compatible endpoint. Set a max-cost budget and Badgr auto-manages provisioning, health checks, logs, and teardown.

Ready to run on L40S 48GB?

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

Back to capacityView docs