Class: PerplexityProvider
Defined in: providers/perplexity.ts:53
Perplexity LLM Provider
Provides streaming chat completion using Perplexity’s API. Sonar models include real-time web search capabilities.
Extends
Constructors
Constructor
new PerplexityProvider(config?): PerplexityProvider;
Defined in: providers/perplexity.ts:58
Parameters
| Parameter | Type |
|---|---|
config | PerplexityProviderConfig |
Returns
PerplexityProvider
Overrides
OpenAICompatibleProvider.constructor
Properties
name
readonly name: "perplexity" = 'perplexity';
Defined in: providers/perplexity.ts:54
Provider name (e.g., ‘openai’, ‘ollama’, ‘gemini’)
Overrides
Methods
chat()
chat(messages, options?): AsyncIterable<StreamChunk>;
Defined in: providers/openai-compatible.ts:222
Stream chat completion from the provider
Parameters
| Parameter | Type | Description |
|---|---|---|
messages | Message[] | Conversation messages |
options? | ChatOptions | Chat options (thinking is ignored for non-Claude providers) |
Returns
AsyncIterable<StreamChunk>
Inherited from
countTokens()
countTokens(messages): Promise<number>;
Defined in: providers/openai-compatible.ts:631
Count tokens in messages using tiktoken (cl100k_base encoding)
Parameters
| Parameter | Type |
|---|---|
messages | Message[] |
Returns
Promise<number>
Inherited from
OpenAICompatibleProvider.countTokens
getModel()
getModel(): string;
Defined in: providers/openai-compatible.ts:158
Get the current default model ID.
Returns
string
Inherited from
OpenAICompatibleProvider.getModel
setModel()
setModel(modelId): void;
Defined in: providers/openai-compatible.ts:162
Change the default model for subsequent calls. Same provider only. Takes effect on the next chat() call, not mid-stream.
Parameters
| Parameter | Type | Description |
|---|---|---|
modelId | string | The new model ID (e.g., ‘claude-opus-4-20250514’) |
Returns
void