How to use Cline with a self-hosted model on a cloud GPU

Rent a GPU on Deploy, then open a terminal in JupyterLab or over SSH and serve an open-weight model there. Install Cline in VS Code the way the official install page prints, set the OpenAI Compatible provider to the forwarded address of that server, and copy anything you saved on the instance before you destroy it. Destroying the instance deletes its disk and stops the server.

An agent can rent the GPU and serve the model the editor calls. The agent pages are GPU compute for AI agents and the agent integration guide. Those pages are how an agent discovers a GPU and runs work on it. This page is the Cline setting. A coding-agent server whose card prints --tool-call-parser qwen3_coder is How to serve Qwen3-Coder for coding agents on a cloud GPU. A terminal client whose OpenAI-compatible page sets OPENAI_API_BASE is How to use Aider with a self-hosted model on a cloud GPU. A local agent UI whose Docker command publishes port 3000 is How to run OpenHands with a self-hosted model on a cloud GPU. A general vLLM install is How to serve an OpenAI-compatible API with vLLM on a cloud GPU. A general SGLang install is How to run SGLang on a cloud GPU. Ollama’s install and port 11434 are How to run Ollama on a cloud GPU. CloudGPU’s hosted gateway is a different base URL: How do I use an OpenAI-compatible API for open models?. The Cursor fields on How do I use CloudGPU as a custom base URL in Cursor? are for that hosted URL, not for the localhost URL of a server you start. How published weight sizes are read: How much VRAM do I need to run an LLM?. If the process runs out of memory: How to fix CUDA out of memory on a cloud GPU. If the server raises no kernel image on an RTX 5090: How to fix CUDA no kernel image on an RTX 5090. The tunnel is How to use SSH port forwarding on a cloud GPU. A second editor whose config.yaml sets an apiBase for chat, autocomplete, and embeddings is How to use Continue with a self-hosted model on a cloud GPU.

The Cline LICENSE file begins with “Apache License Version 2.0, January 2004”. The appendix copyright line is “Copyright 2026 Cline Bot Inc.” The GitHub license field on 2026-11-14 was Apache-2.0. That license text is not a CloudGPU price.

The steps below are copied from the upstream pages named in each step. They were not run on a CloudGPU instance for this page. Anything this page adds around them is marked. The OpenAI Compatible page also names other providers. This page does not list those names.

What the local-models page prints

The local-models page (fetched 2026-11-14) says to run Cline with local inference, and its quick start is: install a local runtime, start the local server, select the matching provider in Cline Settings, select a local model, and enable Use Compact Prompt in Cline Settings, Features.

Its hardware table is labeled RAM, not GPU memory:

RAM Typical local setup
16-32GB Small/quantized models
32-64GB Mid-size coding models
64GB+ Larger models and bigger context windows

The same page’s recommended settings say to enable Use Compact Prompt, keep tasks focused, and start a new task when context gets too large. It does not print a token count. The OpenAI Compatible page lists Max Output Tokens and Context Window size as Model Configuration fields. It does not print a number for either field. This page does not invent one.

The Ollama tab says to select provider Ollama and set Base URL to http://localhost:11434. The LM Studio tab says to select provider LM Studio and keep Base URL as http://localhost:1234. Those two URLs are for a runtime on the same machine as Cline. The same page also documents Atomic Chat. This page does not copy that desktop path.

Choosing a GPU

The local-models page prints the RAM table above and prints no GPU GB figure. The OpenAI Compatible page prints no GPU GB figure. This page does not treat the RAM rows as vram_gb and does not mark a catalog row as enough.

GET /v1/public/compute-capabilities does not list GPUs. It returns feature flags. The memory column is vram_gb from gpu-prices.json, updated_at 2026-11-14T01:00:35.518Z. GET /v1/public/gpu-catalog the same day (generated_at 2026-11-14T01:00:36.104Z) listed the same names and the same vram_gb values, plus Ascend 910B at 64, which is not in gpu-prices.json and is not listed here. The catalog response also includes an availability field. This page does not copy it. Every prices row has gpu_count 1. Every catalog row has min_gpu_count 1. Which listing you can start is on Deploy.

The catalog template list had no Cline template. It did have Ollama templates. Those templates are a different install. This page does not treat them as Cline.

Catalog name vram_gb
RTX 2080 8
RTX 3080 10
RTX 2080 Ti 11
RTX 3080 Ti 12
V100 16G 16
A30 24
RTX 3090 24
RTX 4090 24
RTX 4090D 24
RTX 4080 SUPER 32G 32
RTX 5090 32
RTX 5090D 32
V100 32G 32
A100 40G 40
A800 40G 40
A40 48
RTX 4090 48G 48
RTX 4090D 48G 48
RTX A6000 48
NVIDIA RTX PRO 5000 72G 72
RTX PRO 5000 72G 72
A100 80G 80
A800 80G 80
H800 80
RTX 6000D 83
H20 96G 96
RTX PRO 6000 96G 96
RTX PRO 6000 Max-Q 96
H20-3e 141G 141

Install Cline in VS Code

The install page’s IDE Extension steps for VS Code, copied. The page says to press Ctrl/Cmd + Shift + X, type Cline, click Install, then find Cline in the extension sidebar.

Cline itself is the editor extension. The weights are whatever model server you start on the instance. This page does not clone the Cline repository onto the GPU and does not add a weight download. A vLLM or SGLang server is the two guides linked above. An Ollama server is the Ollama guide.

The same install page also prints a CLI path: Node.js 20 or newer, with 22 called recommended, then:

npm install -g cline

Then cline auth and cline. This page follows the VS Code steps. It did not run the CLI.

Point Cline at the server

The OpenAI Compatible page’s general settings, limited to the fields this page keeps. It says to open the Cline settings panel and set:

  • API Provider: “OpenAI Compatible”.
  • Base URL: the endpoint of the server you started.
  • API Key: the key that server expects.
  • Model: the model id that server serves.

The same section also lists price fields. This page does not copy them.

For a vLLM or SGLang process on the instance, the Base URL is the client address that server’s guide prints, after the port is forwarded. How to use SSH port forwarding on a cloud GPU is how a program on your computer reaches a port on the instance. This page does not add an ssh command. The hosted gateway at https://cloudgpu.app/v1 is the other guide, not this server.

The local-models page’s Ollama Base URL is http://localhost:11434, and its LM Studio Base URL is http://localhost:1234. Use those when that runtime is on the same machine as Cline. They are not a substitute for the forwarded address of a server that is bound on the instance.

When it fails

CUDA out of memory. The local-models page prints RAM rows and no GPU GB figure. If the model server raises CUDA out of memory, use how to fix CUDA out of memory on a cloud GPU. This page does not pick a smaller model id for you.

“CUDA error: no kernel image is available for execution on the device” on an RTX 5090. The Cline pages this page read do not pin a PyTorch build and do not name compute capability 12.0. The server you start might. The RTX 5090 guide says that card is compute capability 12.0 (sm_120) and that PyTorch 2.7 and later, on a build for CUDA 12.8 or newer, is the line that announced support. This page did not install a torch wheel and did not re-test a server on an RTX 5090. Catalog RTX 5090 and RTX 5090D are 32. That is the catalog number only.

The OpenAI Compatible page’s troubleshooting list names “Invalid API Key”, “Model Not Found”, and connection errors. It says to check the API key, the model id, and the Base URL. This page did not reproduce those errors.

What to copy off, then destroy

The Cline pages do not name an output directory on the GPU. Chat transcripts stay in the editor. Any file the agent writes on the instance is on that disk. Copy it before you destroy the instance. Save your work before you destroy a cloud GPU is the checklist. The terms say data on local instance storage is permanently deleted when the instance is destroyed. The server stops with the instance.

The pricing page FAQ states the billing rule in these words:

Per-minute options have a 1-minute minimum; unused prepaid minutes are refunded when you destroy the instance. Some configurations use whole-hour billing with a 1-hour minimum; unused time within a charged hour is not refunded on voluntary stop. The billing period, minimum charge and initial prepayment are shown before confirmation. No subscription.

Prices are not copied here.

Frequently asked questions

How much memory does the local-models page print?

The local-models page prints a RAM table: 16-32GB for small or quantized models, 32-64GB for mid-size coding models, and 64GB+ for larger models and bigger context windows. It prints no GPU GB figure. The OpenAI Compatible page lists Context Window size and Max Output Tokens as fields and prints no token count. This page does not mark a catalog row as enough. Which listing you can start is on https://cloudgpu.app/deploy.

Which install and provider settings do the Cline docs print?

The VS Code steps on the install page are Open Extensions, search Cline, Install, then open Cline. The OpenAI Compatible page says to select API Provider OpenAI Compatible and set Base URL, API Key, and Model ID. The local-models page says to select provider Ollama with Base URL http://localhost:11434, or provider LM Studio with Base URL http://localhost:1234, and to enable Use Compact Prompt. This page did not save those settings.

How does Cline reach a model on the rented GPU?

The documented VS Code install runs in the editor. The model server is the process you start on the instance. The base URLs on the local-models page are for a runtime on the same machine as Cline. A vLLM or SGLang server on the instance is reached through the SSH port-forwarding guide, and the OpenAI Compatible Base URL is the address that server guide prints. This page did not open a tunnel.

What license does the file print?

The Cline LICENSE file begins with Apache License Version 2.0, January 2004. The appendix copyright line is Copyright 2026 Cline Bot Inc. The GitHub license field on 2026-11-14 was Apache-2.0. That license text is not a CloudGPU price.

Sources (read 2026-11-14, Asia/Shanghai)

Cards you can start are on the Deploy page. Prices are on https://cloudgpu.app/pricing and are not repeated here.

Last updated: 2026-11-14 (Asia/Shanghai).