Class OpenRouterClient

Chat Completions

batchChatCompletion

batchChatCompletion(
requests: ChatCompletionParams[],
options?: BatchOptions,
): Promise<BatchResult[]>

Executes multiple chat completion requests in parallel (batch) Automatically handles rate limiting and individual errors

Parameters

requests: ChatCompletionParams[]
Array of request parameters
options: BatchOptions = {}
Batch options (maxConcurrent, stopOnError)

Returns Promise<BatchResult[]>

Results of each request (success or error)

Example

const results = await client.batchChatCompletion([
  { model: 'openai/gpt-3.5-turbo', messages: [{ role: 'user', content: 'Q1' }] },
  { model: 'openai/gpt-3.5-turbo', messages: [{ role: 'user', content: 'Q2' }] }
], { maxConcurrent: 5, stopOnError: false });

results.forEach((result, idx) => {
  if (result.success && result.response) {
    console.log(`Request ${idx}: ${result.response.choices[0].message.content}`);
  } else {
    console.error(`Request ${idx} failed: ${result.error?.message}`);
  }
});

Validation & Helpers

supportsFeature

supportsFeature(modelId: string, feature: ModelFeature): Promise<boolean>
Quickly checks if a model supports a specific feature
Parameters
- modelId: string
  The model ID to check
- feature: ModelFeature
  The feature to check ('streaming', 'tools', 'vision', 'json')
Returns Promise<boolean>
True if the model supports the feature
Example
```
const supportsVision = await client.supportsFeature('anthropic/claude-3.5-sonnet', 'vision');
if (supportsVision) {
  // Send an image
}
```
- Defined in openrouter.ts:1327

validateParams

validateParams(
    modelId: string,
    params: {
        messages?: ChatMessage[];
        stream?: boolean;
        tools?: any[];
        response_format?: any;
        max_tokens?: number;
        [key: string]: any;
    },
): Promise<ValidationResult>
Validates request parameters against model capabilities
Parameters
- modelId: string
  The model ID
- params: {
      messages?: ChatMessage[];
      stream?: boolean;
      tools?: any[];
      response_format?: any;
      max_tokens?: number;
      [key: string]: any;
  }
  The parameters to validate
Returns Promise<ValidationResult>
Validation result with errors and warnings
Example
```
const validation = await client.validateParams('openai/gpt-3.5-turbo', {
  messages: [{ role: 'user', content: 'Hello' }],
  stream: true,
  max_tokens: 5000
});

if (!validation.valid) {
  console.error('Errors:', validation.errors);
}
if (validation.warnings?.length) {
  console.warn('Warnings:', validation.warnings);
}
```
- Defined in openrouter.ts:1374

truncateMessages

truncateMessages(messages: ChatMessage[], maxTokens: number): ChatMessage[]
Intelligently truncates a list of messages to respect a token limit Always preserves the system message if it exists, and truncates from the beginning (FIFO)
Parameters
- messages: ChatMessage[]
  The messages to truncate
- maxTokens: number
  Maximum token limit (approximate, based on 4 chars ≈ 1 token)
Returns ChatMessage[]
The truncated messages
Example
```
const longConversation = [
  { role: 'system', content: 'You are helpful' },
  // ... 50 messages
];

const truncated = client.truncateMessages(longConversation, 4000);
// Keeps the system message + the most recent messages that fit within 4000 tokens
```
- Defined in openrouter.ts:1488

Other

constructor

new OpenRouterClient(config: OpenRouterConfig): OpenRouterClient
Creates an instance of the OpenRouter client.
Parameters
- config: OpenRouterConfig
  Client configuration.
  - apiKey: string
  - OptionalbaseURL?: string
  - Optionaltimeout?: number
  - OptionalmaxRetries?: number
  - OptionalretryDelay?: number
  - Optionalheaders?: Record<string, string>
  - Optionallogger?: Logger
  - OptionallogLevel?: LogLevel
Returns OpenRouterClient
- Defined in openrouter.ts:536

createChatCompletion

createChatCompletion(
params: ChatCompletionParams,
): Promise<ChatCompletionResponse>
Sends a chat completion request to the OpenRouter API.
Parameters
- params: ChatCompletionParams
  Complete request parameters (model, messages, and all optional parameters).
Returns Promise<ChatCompletionResponse>
The API response.
- Defined in openrouter.ts:789

createChatCompletionStream

createChatCompletionStream(
params: ChatCompletionParams,
): Promise<ReadableStream<any>>
Sends a chat completion request with streaming to the OpenRouter API.
Parameters
- params: ChatCompletionParams
  Complete request parameters (model, messages, and all optional parameters).
Returns Promise<ReadableStream<any>>
Response stream.
- Defined in openrouter.ts:831

streamChatCompletion

streamChatCompletion(
params: ChatCompletionParams,
): AsyncIterable<ChatCompletionChunk>
Sends a streaming chat completion request and returns an AsyncIterable. This method provides a more ergonomic API than createChatCompletionStream by converting the ReadableStream to AsyncIterable, allowing the use of for await...of.
Parameters
- params: ChatCompletionParams
  Complete request parameters (model, messages, and all optional parameters).
Returns AsyncIterable<ChatCompletionChunk>
AsyncIterable of response chunks.
Example
```
for await (const chunk of client.streamChatCompletion({
  model: 'openai/gpt-3.5-turbo',
  messages: [{ role: 'user', content: 'Hello!' }]
})) {
  console.log(chunk.choices[0]?.delta?.content || '');
}
```
- Defined in openrouter.ts:885

countTokens

countTokens(text: string): number
Counts approximately the number of tokens in a text
Parameters
- text: string
  Text to analyze
Returns number
Approximate number of tokens
- Defined in openrouter.ts:1085

countMessagesTokens

countMessagesTokens(messages: ChatMessage[]): number
Counts tokens in an array of messages
Parameters
- messages: ChatMessage[]
  Messages to analyze
Returns number
Total approximate number of tokens
- Defined in openrouter.ts:1095

listModels

listModels(): Promise<ModelInfo[]>
Retrieves the list of available models

Returns Promise<ModelInfo[]>
List of available models
- Defined in openrouter.ts:1138

getModel

getModel(modelId: string): Promise<ModelInfo>
Retrieves information for a specific model
Parameters
- modelId: string
  Model ID (format: "author/name")
Returns Promise<ModelInfo>
Model information
- Defined in openrouter.ts:1167

getModelEndpoints

getModelEndpoints(modelId: string): Promise<ModelEndpoint[]>
Retrieves available endpoints for a specific model
Parameters
- modelId: string
  Model ID (format: "author/name")
Returns Promise<ModelEndpoint[]>
List of model endpoints
- Defined in openrouter.ts:1192

getModelCapabilities

getModelCapabilities(modelId: string): Promise<ModelCapabilities>
Retrieves complete model capabilities (supported parameters, modalities, limits) Combines data from getModel() and getModelEndpoints() for an overview
Parameters
- modelId: string
  Model ID (e.g: 'anthropic/claude-3.5-sonnet')
Returns Promise<ModelCapabilities>
Detailed model capabilities
Example
```
const caps = await client.getModelCapabilities('anthropic/claude-3.5-sonnet');
if (caps.supportsTools) {
  // Use function calling
}
if (caps.supportsVision) {
  // Send images
}
```
- Defined in openrouter.ts:1240

validateApiKey

validateApiKey(): Promise<boolean>
Checks if an API key is valid

Returns Promise<boolean>
True if the key is valid
- Defined in openrouter.ts:1560

getGeneration

getGeneration(generationId: string): Promise<GenerationStats>
Retrieves statistics for a specific generation
Parameters
- generationId: string
  The generation ID
Returns Promise<GenerationStats>
The generation statistics
- Defined in openrouter.ts:1578

listProviders

listProviders(): Promise<Provider[]>
Retrieves the list of available providers

Returns Promise<Provider[]>
List of available providers
- Defined in openrouter.ts:1616

getCredits

getCredits(): Promise<Credits>
Retrieves account credits

Returns Promise<Credits>
Account credits (total_credits and total_usage)
- Defined in openrouter.ts:1644

getKeyInfo

getKeyInfo(): Promise<KeyInfo>

Retrieves detailed information for the API key (usage, limits, rate limits) Endpoint: GET /auth/key

Returns Promise<KeyInfo>

Information about the API key

Example

const keyInfo = await client.getKeyInfo();
console.log('Usage:', keyInfo.usage);
console.log('Limit:', keyInfo.limit || 'Unlimited');
if (keyInfo.rate_limit) {
  console.log('Rate limit:', keyInfo.rate_limit.requests, 'per', keyInfo.rate_limit.interval);
}

getActivity

getActivity(): Promise<ActivityStats>
Retrieves account activity statistics

Returns Promise<ActivityStats>
Activity statistics
- Defined in openrouter.ts:1750

`Static`parseToolCalls

parseToolCalls(toolCalls: ToolCall[]): ParsedToolCall[]
Parses tool_calls from a response and returns parsed arguments
Parameters
- toolCalls: ToolCall[]
  Tool calls to parse
Returns ParsedToolCall[]
Tool calls with parsed arguments
- Defined in openrouter.ts:1774

`Static`createToolResponseMessage

createToolResponseMessage(
    toolCallId: string,
    content: string,
    name?: string,
): ChatMessage
Creates a message of type 'tool' to respond to a tool call
Parameters
- toolCallId: string
  Tool call ID
- content: string
  Response content (can be a JSON stringified)
- Optionalname: string
  Function name (optional)
Returns ChatMessage
Formatted message for the API
- Defined in openrouter.ts:1805

`Static`createToolResponseFromResult

createToolResponseFromResult(
    toolCallId: string,
    result: any,
    name?: string,
): ChatMessage
Helper to create a tool response message from any object result. Automatically serializes objects to JSON string.
Parameters
- toolCallId: string
  The ID of the tool call this is responding to
- result: any
  The result object or string to include as content
- Optionalname: string
  Optional name of the tool function
Returns ChatMessage
A properly formatted ChatMessage for tool responses
Example
```
const weatherResult = { location: 'Paris', temperature: 22, conditions: 'Sunny' };
const message = OpenRouterClient.createToolResponseFromResult(
  'call_123', 
  weatherResult, 
  'get_weather'
);
```
- Defined in openrouter.ts:1837

`Static`markMessageAsCacheable

markMessageAsCacheable(message: ChatMessage): ChatMessage
Marks a message as cacheable for Anthropic Prompt Caching. Adds a "cache breakpoint" allowing to cache all prompt content up to this point.

Savings: 90% reduction on cached tokens (hit), 10% surcharge (miss). Ideal for: long system prompt, repeated context, multi-turn conversations.

⚠️ Note: OpenRouter does not yet return cache metrics (cache_read_input_tokens, cache_creation_input_tokens) even though caching works on Anthropic's side. You will save money but won't see detailed cache metrics in API responses.
Parameters
- message: ChatMessage
  Message to mark as cacheable
Returns ChatMessage
New message with cache_control added
Example
```
const systemPrompt = OpenRouterClient.markMessageAsCacheable({
  role: 'system',
  content: 'Long system prompt...'
});
```
- Defined in openrouter.ts:1866

`Static`executeToolCalls

executeToolCalls(
toolCalls: ToolCall[],
functions: Record<string, (...args: any[]) => any | Promise<any>>,
): Promise<ChatMessage[]>
Helper to execute tool calls and create response messages
Parameters
- toolCalls: ToolCall[]
  Tool calls to execute
- functions: Record<string, (...args: any[]) => any | Promise<any>>
  Map of available functions
Returns Promise<ChatMessage[]>
Formatted response messages
- Defined in openrouter.ts:1909

Class OpenRouterClient

Index

Chat Completions

Validation & Helpers

Other

Chat Completions

batchChatCompletion

Parameters

Returns Promise<BatchResult[]>

Example

Validation & Helpers

supportsFeature

Parameters

Returns Promise<boolean>

Example

validateParams

Parameters

Returns Promise<ValidationResult>

Example

truncateMessages

Parameters

Returns ChatMessage[]

Example

Other

constructor

Parameters

apiKey: string

OptionalbaseURL?: string

Optionaltimeout?: number

OptionalmaxRetries?: number

OptionalretryDelay?: number

Optionalheaders?: Record<string, string>

Optionallogger?: Logger

OptionallogLevel?: LogLevel

Returns OpenRouterClient

createChatCompletion

Parameters

Returns Promise<ChatCompletionResponse>

createChatCompletionStream

Parameters

Returns Promise<ReadableStream<any>>

streamChatCompletion

Parameters

Returns AsyncIterable<ChatCompletionChunk>

Example

countTokens

Parameters

Returns number

countMessagesTokens

Parameters

Returns number

listModels

Returns Promise<ModelInfo[]>

getModel

Parameters

Returns Promise<ModelInfo>

getModelEndpoints

Parameters

Returns Promise<ModelEndpoint[]>

getModelCapabilities

Parameters

Returns Promise<ModelCapabilities>

Example

validateApiKey

Returns Promise<boolean>

getGeneration

Parameters

Returns Promise<GenerationStats>

listProviders

Returns Promise<Provider[]>

getCredits

Returns Promise<Credits>

getKeyInfo

Returns Promise<KeyInfo>

Example

getActivity

Returns Promise<ActivityStats>

StaticparseToolCalls

Parameters

Returns ParsedToolCall[]

`Optional`baseURL?: string

`Optional`timeout?: number

`Optional`maxRetries?: number

`Optional`retryDelay?: number

`Optional`headers?: Record<string, string>

`Optional`logger?: Logger

`Optional`logLevel?: LogLevel

`Static`parseToolCalls

`Static`createToolResponseMessage

`Static`createToolResponseFromResult

`Static`markMessageAsCacheable

`Static`executeToolCalls