Three routes
| Route | Largest model | Context | You pay | You need |
|---|---|---|---|---|
| Chat app with an API key | Kimi K3.1 Pro, 2,780B | 1,048,576 | Per token, credits from $5 | An API key |
| Coding agent | Kimi K3.1 Pro, 2,780B | 1,048,576 | Per token, or a plan from $10 a month | An API key and the agent |
| A local GPU | Qwen 3.6 35B-A3B Abliterated, 36B | 262,144 | The hardware | One 24 GB GPU |
Set one up
Chat app
- Open the app's model settings, for example Open WebUI.
- Pick OpenAI Compatible as the provider.
- Set the base URL to
https://api.silkcompute.com. - Paste your key and a model id such as
kimi-k3.1-pro-uncensored.
Coding agent
- Claude Code reads
ANTHROPIC_BASE_URL. - Set it to
https://api.silkcompute.com. - Codex and OpenCode take the same base URL in their config.
A local GPU
- Install Ollama from ollama.com.
- Run one command:
ollama run hf.co/mlabonne/Meta-Llama-3.1-8B-Instruct-abliterated-GGUF
Limits that remain
Silk Compute's Terms forbid sexual content involving anyone under 18. Refusal behavior is removed, and models are served without content filters.