basalt / ai/src / OpenAICompatibleProvider
Class: OpenAICompatibleProvider
Defined in: ai/src/provider/openai.ts:48
Provider for any OpenAI-compatible Chat Completions endpoint (OpenAI itself, Azure-style gateways, LiteLLM, OpenRouter, go4ai, …). Unlike Anthropic's API, system messages stay inline in the messages array.
Streams by default — the robust path for large generations behind a gateway.
Implements
Constructors
Constructor
> new OpenAICompatibleProvider(options): OpenAICompatibleProvider
Defined in: ai/src/provider/openai.ts:57
Parameters
options
OpenAICompatibleProviderOptions
Returns
OpenAICompatibleProvider
Properties
model
> readonly model: string
Defined in: ai/src/provider/openai.ts:50
Concrete model id in use, e.g. 'claude-sonnet-5'.
Implementation of
name
> readonly name: "openai" = 'openai'
Defined in: ai/src/provider/openai.ts:49
Stable id, e.g. 'anthropic'.
Implementation of
Methods
generate()
> generate(options): Promise<string>
Defined in: ai/src/provider/openai.ts:69
One-shot completion — returns the assistant's full text.
Parameters
options
Returns
Promise<string>
Implementation of
stream()
> stream(options): AsyncIterable<string>
Defined in: ai/src/provider/openai.ts:76
Streaming completion — yields chunks as they arrive.
Parameters
options
Returns
AsyncIterable<string>