@makaio/provider-nanogpt
v1.0.0-dev-1779051654000
Published
Type-only provider package for NanoGPT — a specialized OpenAI-compatible inference gateway. This package declares the provider identity, endpoint URL, credential environment variable, and filter mode. It contains no runtime logic; the model catalog is pop
Readme
@makaio/provider-nanogpt
Type-only provider package for NanoGPT — a specialized OpenAI-compatible inference gateway. This package declares the provider identity, endpoint URL, credential environment variable, and filter mode. It contains no runtime logic; the model catalog is populated from the YAML lab registry at boot time. Because NanoGPT is a specialized provider, defaultModelFilterMode is set to 'allowlist' — models are hidden by default and must be explicitly allowed. The OpenAI Node adapter serves this provider.
Provider Identity
| Field | Value |
|-------|-------|
| id | nanogpt |
| name | NanoGPT |
| protocol | openai |
| defaultModel | — (none; allowlist mode) |
| fastModel | — (none; allowlist mode) |
| defaultModelFilterMode | allowlist |
Credential Environment Variables
| Variable | Description |
|----------|-------------|
| NANOGPT_API_KEY | API key for the NanoGPT inference API |
Endpoints
| Protocol | URL |
|----------|-----|
| openai | https://nano-gpt.com/api/v1 |
Served By
@makaio/ai-adapters-openai-node— drives NanoGPT via the OpenAI Chat Completions wire protocol.
Installation
@makaio/provider-nanogpt is a private workspace package used internally by the framework.
