Running in ~30 s
Pre-cached image, ports mapped, TLS terminated — the stack is working before you finish reading this.
Template · Training & fine-tuning · CUDA 12.9
Run Unsloth fine-tuning on a cloud GPU, in 30 seconds
2× faster, lower-VRAM fine-tuning — Unsloth kernels with a notebook workflow. Unsloth rewrites the attention and MLP kernels so LoRA/QLoRA fine-tuning runs about 2× faster with up to 70% less VRAM. Unsloth Studio adds a notebook workflow on top: load a base model, attach a dataset, train, export GGUF or merged weights. From $0.037/hr on an interruptible RTX 4060 Ti.
Template is free
You pay the GPU price only: Unsloth Studio on a RTX 4060 Ti is $0.074/hr on-demand, $0.037/hr interruptible, billed per second.
Volumes for state
Models, datasets and outputs live on a $0.08/GB/mo volume; the instance stays disposable.
Private by default
Dedicated GPU, encrypted disk, crypto payments, no KYC, no stored IPs — Jupyter and SSH behind your own credentials.
Best GPUs for Unsloth Studio
Three price points that run this template well — Good, Better, Best. Every model in the catalogue works; these are the value picks.
| Tier | GPU | VRAM | On-demand | Interruptible | Why this card | Action |
|---|---|---|---|---|---|---|
| Good | RTX 4060 Ti | 8 GB | $0.074 | $0.037 | 16 GB is enough for 8B QLoRA with Unsloth's memory savings. | Deploy |
| Better | RTX 4090 | 24 GB | $0.262 | $0.131 | 24 GB handles 14B–32B QLoRA quickly. | Deploy |
| Best | RTX 5090 | 32 GB | $0.318 | $0.159 | 32 GB GDDR7 for the fastest single-card fine-tunes. | Deploy |
Need more VRAM? The full catalogue lists all 76 models with live availability; the VRAM guide sizes models to cards.
Deploy Unsloth Studio from the console, CLI or API
Pick the template in the console deploy bar, or script it:
- Console — filter by GPU, choose Unsloth Studio in the template picker, set disk and env, deploy.
- CLI — pip install powergpu, then the command on the right. CLI reference.
- API — POST /v1/instances with "template": "unsloth-studio". REST reference.
- Own image — any OCI reference works too; we inject the NVIDIA runtime. Template docs.
$ powergpu launch --gpu rtx-4060-ti --template unsloth-studio \
--disk 100 --volume models:/workspace/models
✓ instance i-7a41c0e2 running (27.9s)
# Unsloth Studio · RTX 4060 Ti · $0.074/hr · per second
# https://i-7a41c0e2.powergpu.io:8888 (TLS)
$ powergpu stop i-7a41c0e2 # billing ends this secondWhat is inside the Unsloth Studio template
| Image | powergpu/unsloth-studio |
|---|---|
| CUDA | CUDA 12.9 |
| Access | SSH shell · JupyterLab on a mapped port |
| Category | Training & fine-tuning |
| Storage | Instance NVMe disk (sized at deploy) + optional network volumes |
| Billing | GPU price only, per second — no template fee, no setup fee |
Other training & fine-tuning templates
Axolotl — Fine Tuning on a cloud GPU — Fine-tune Llama, Qwen or Mistral from one YAML — LoRA, QLoRA and multi-GPU FSDP.
Kohya's GUI on a cloud GPU — Train SDXL and Flux image LoRAs in a browser GUI — dataset in, .safetensors out.
Flux Gym on a cloud GPU — A dead-simple UI for training Flux LoRAs, built on the Kohya scripts.
Ostris AI Toolkit on a cloud GPU — The AI-Toolkit trainer for Flux and diffusion models, with a web UI.- All 37 templates
Unsloth Studio on a cloud GPU: FAQ
Environment variables, ports and custom images are covered in the template docs.
How much does it cost to run Unsloth Studio on a cloud GPU?
Only the GPU price — the template is free. From $0.037/hr on an interruptible RTX 4060 Ti, $0.074/hr on-demand on a RTX 4060 Ti. Billing is per second, so an hour of tinkering costs an hour, not a day. Storage is $0.08/GB/month.
How long does Unsloth Studio take to start?
About 30 seconds from the deploy click: the image is pre-cached on hosts, ports are mapped and TLS is terminated for you. Restarting a stopped instance is faster, and your disk is exactly as you left it.
Can I keep my models and outputs between sessions?
Yes — attach a volume at deploy. Everything on the volume survives instance destruction and mounts on the next instance in the region in seconds, at $0.08/GB/month. The instance disk itself survives stop/start but not destroy.
Which models does Unsloth support?
Llama 3.x, Qwen 2.5/3, Gemma, Mistral, Phi and most decoder-only architectures on Hugging Face, plus vision-language variants in recent releases.