iwant: launch an open model with one command
iwant turns a documented model name into a cloud endpoint: it provisions the GPU host, configures serving, checks readiness, and prints a URL and key. You do not assemble recipes, Docker images, or model-specific engine flags yourself.
iwant up gpt-oss-20bGCP is currently supported. Complete installation and cloud authentication first; your account needs billing, GPU quota, and capacity.
When to choose iwant
| Signal | What it means |
|---|---|
| You know the model name | iwant up MODEL selects serving configuration for documented checkpoints |
| You want one teardown command | iwant down removes the cluster — see billing and teardown |
| You prefer ops automation over engine tuning | Under the hood, documented models use vLLM; you still validate quality and concurrency |
Choose vLLM or SGLang when you need custom hardware layouts, parsers, or flags iwant does not expose.
Setup and operation
| Guide | What you’ll learn |
|---|---|
| Install and cloud setup | GCP auth, dry-run preview, first launch, API check, teardown |
| Alternatives to iwant | When to run vLLM, SGLang, or local runtimes instead |
Preview cloud resources without renting a GPU:
iwant up gpt-oss-20b --infra gcp --dry-run --yesThen follow a model deployment guide for hardware-specific notes and the first inference request.
Model-specific deployment guides
Each guide links the actual upstream configuration, explains hardware and checkpoint variants, and includes a first API request. Browse all documented models for other deployment options.
iwant alternatives
| Tool | Choose when |
|---|---|
| vLLM | You manage the GPU host and need continuous batching, prefix caching, and direct control over vllm serve settings. |
| SGLang | You need RadixAttention prefix reuse, structured generation, or model-specific parsers on hardware you configure. |
| TGI | You already run Hugging Face Text Generation Inference on Kubernetes and the checkpoint is on the supported-model list. |
| Ollama | You want local ollama pull ergonomics, IDE chat, and low-ops serving on a single machine. |
| llama.cpp | You serve GGUF weights on CPU, Apple Silicon, or a modest GPU with minimal dependencies. |
| LocalAI | You need one OpenAI-compatible URL over multiple backends or gallery configurations. |
Deeper comparison: Alternatives to iwant. All launch methods: self-hosting hub.
Related reading
- Self-hosting hub — economics, model directory, and launch comparison
- Text generation inference — runtime terminology and workload fit
- Local inference practice — when to self-host and how to meter GPU time
- Alternatives to iwant
- Install and cloud setup