online gpu rental services -- for AI -- with free docker images for salad cheap user rented GPUs

 RunPod -- GPUS with pre installed configured MOdels -- and 

cheap for running model per sec pulse -- 

https://salad.com/pricing -- very cheap --

https://cloud.vast.ai/

Why salad so cheap:

We utilize unused, latent compute resources from gamers and e-sports arenas worldwide to power our network — think Airbnb for GPUs.

• RunPod = real datacenters in US/EU - reliable, fast.

• Salad = thousands of gamer PCs worldwide sleeping at night - they rent their GPU when not gaming. That's why it's cheap.

• Salad good: Insanely cheap ($0.02/hr for RTX 5090), 191 countries

• Salad bad: Distributed = unpredictable availability, Max 24GB VRAM per GPU, Not for low-latency workloads

• RunPod: Per-second billing
• Salad: You only pay for time hardware is available, they don't charge for cold boot. Hourly calculator, but billed per second.
• Both per hour, not per month, but you can keep running 720 hrs = monthly.

Is Salad cheapest?
For batch jobs / non-critical - Yes, Salad is cheapest in world at RTX 3090 $0.09/hr, RTX 4090 $0.16/hr.
For interactive video work where you need GPU NOW - RunPod/Vast.ai better. Salad can be offline when you need it.

2. What happens if you take Salad vs RunPod?

Salad is Airbnb for gamer PCs.

• RunPod: Datacenter GPU, you rent same machine. You have Network Volume - install once, stays forever. You can keep it for months, stop/start.
• Salad: Every time you request, you get a different random gamer's PC somewhere in the world. You cannot keep it long time, max 24-48 hrs, and no Network Volume. You have to use Docker - you build your image with Python + models baked in, push to Salad. Each new PC downloads your Docker - 2-3 min load. If that gamer turns off PC, your job moves to another PC - interruption.

For your pattern of 20 times per day for 3 mins, Salad is actually okay because it's cheap and stateless, but for long 1-hour video work, RunPod is more stable.

Install every time?

• RunPod Pods: No, with Network Volume install once.
• Salad & RunPod Serverless: You bake models into Docker image once, then every launch loads from Docker - 1-2 min, no manual pip install each time.

Time for First Video 

3. Vast.ai - 30 to 60 minutes first time

Setup time: ∼10 minutes (more configuration)

• Signup + credit card verification: 5 mins [they verify ID sometimes]
• Search RTX 3090 $0.11/hr -> Rent -> Choose template: 2 mins
• Machine boots + you have to manually install ComfyUI + LTX via terminal if no template: 15-25 mins
• Generate: 1 min

First video total: 40-60 mins for beginner because UI is confusing.

4. Salad - 1 to 3 hours first time [hardest for beginner]

Setup time: ∼20 minutes (container-based)

Salad doesn't let you just click a template. You have to:

• Signup: 1 min
• You must build a Docker image with your code + push to Docker Hub: 30-60 mins if you never did Docker before
• Then deploy that Docker to Salad: 5 mins download
• Generate: 1 min

First video total: 1.5 to 3 hours first time. That's why it's cheap - they make YOU do the work.

free docker images for SAlad and vast -- 

2. Docker Hub ready images for Salad / Vast:

• f4k3r22/aquiles-video - Video models including LTX, Wan
• notrius/lora-pilot:stable - This one supports SD1, SDXL, FLUX, LTX/LTX2, HunyuanVideo, Wan2.1/Wan2.2 in one image
• georgedavila/ltx-img2vid - LTX image-to-video on Replicate, but Docker pull available

You just paste image name notrius/lora-pilot:stable in Vast.ai / Salad custom image field -> it boots with everything.

3. GitHub one-click:

• lucidprogrammer/wan-video - RunPod Serverless Deployment for Wan 2.1 - already has handler.py + Dockerfile - you just docker push and it works.

So how to save time for your first video:

Don't install Python packages yourself.

RunPod way - 3 clicks:
Templates -> Search LTX -> Deploy 4090 -> Wait 3 mins -> You have ComfyUI with LTX nodes ready -> Generate.

Vast/Salad way:
When creating instance, instead of ubuntu image, paste: f4k3r22/aquiles-video:latest or notrius/lora-pilot:stable -> Boot -> It already has all models.

You will have first video in 10 mins, not 3 hours.

For your RTX 2050 local - use same Docker Desktop - pull notrius/lora-pilot:stable and it will run SD 1.5 on 4GB VRAM.

• f you want cheapest possible and don't mind your 1-min video failing once and restarting - Salad $0.09/hr wins by 2 cents.
• If you want cheap + you can actually find the GPU when you need it at 2am - Vast.ai wins.

so user may pull pc any time so generate video by video -- 

30 sec time limot to pull

so 30 sec is there -- so if i generate by max sec per video and chec if request is there stop else go to next part