Loading API documentation
Run any LLM (Large Language Model) with fal, powered by OpenRouter.
The client provides a convenient way to interact with the model API.
npm install --save @fal-ai/clientSet FAL_KEY as an environment variable in your runtime.
export FAL_KEY="YOUR_API_KEY"The client API handles the API submit protocol. It will handle the request status updates and return the result when the request is completed.
import { fal } from "@fal-ai/client";
const result = await fal.subscribe("openrouter/router/enterprise", {
input: {
prompt: "Write a short story (under 200 words) about an AI that learns to dream. Use vivid sensory details and end with a surprising twist that makes the reader feel both awe and melancholy.",
model: "anthropic/claude-opus-5.5"
},
logs: true,
onQueueUpdate: (update) => {
if (update.status === "IN_PROGRESS") {
update.logs.map((log) => log.message).forEach(console.log);
}
},
});
console.log(result.data);
console.log(result.requestId);This model supports streaming requests. You can stream data directly to the model and get the result in real-time.
import { fal } from "@fal-ai/client";
const stream = await fal.stream("openrouter/router/enterprise", {
input: {
prompt: "Write a short story (under 200 words) about an AI that learns to dream. Use vivid sensory details and end with a surprising twist that makes the reader feel both awe and melancholy.",
model: "anthropic/claude-opus-5.5"
}
});
for await (const event of stream) {
console.log(event);
}
const result = await stream.done();The API uses an API Key for authentication. It is recommended you set the FAL_KEY environment variable in your runtime when possible.
import { fal } from "@fal-ai/client";
fal.config({
credentials: "YOUR_FAL_KEY"
});The client API provides a convenient way to submit requests to the model.
import { fal } from "@fal-ai/client";
const { request_id } = await fal.queue.submit("openrouter/router/enterprise", {
input: {
prompt: "Write a short story (under 200 words) about an AI that learns to dream. Use vivid sensory details and end with a surprising twist that makes the reader feel both awe and melancholy.",
model: "anthropic/claude-opus-5.5"
},
webhookUrl: "https://optional.webhook.url/for/results",
});You can fetch the status of a request to check if it is completed or still in progress.
import { fal } from "@fal-ai/client";
const status = await fal.queue.status("openrouter/router/enterprise", {
requestId: "764cabcf-b745-4b3e-ae38-1200304cf45b",
logs: true,
});Once the request is completed, you can fetch the result. See the Output Schema for the expected result format.
import { fal } from "@fal-ai/client";
const result = await fal.queue.result("openrouter/router/enterprise", {
requestId: "764cabcf-b745-4b3e-ae38-1200304cf45b"
});
console.log(result.data);
console.log(result.requestId);Some attributes in the API accept file URLs as input. Whenever that's the case you can pass your own URL or a Base64 data URI.
You can pass a Base64 data URI as a file input. The API will handle the file decoding for you. Keep in mind that for large files, this alternative although convenient can impact the request performance.
You can also pass your own URLs as long as they are publicly accessible. Be aware that some hosts might block cross-site requests, rate-limit, or consider the request as a bot.
We provide a convenient file storage that allows you to upload files and use them in your requests. You can upload files using the client API and use the returned URL in your requests.
import { fal } from "@fal-ai/client";
const file = new File(["Hello, World!"], "hello.txt", { type: "text/plain" });
const url = await fal.storage.upload(file);Read more about file handling in our file upload guide.
prompt string* requiredPrompt to be used for the chat completion
system_prompt stringSystem prompt to provide context or instructions to the model
model string* requiredName of the model to use. Charged based on actual token usage.
reasoning booleanShould reasoning be the part of the final answer.
temperature floatThis setting influences the variety in the model's responses. Lower values lead to more predictable and typical responses, while higher values encourage more diverse and less common responses. At 0, the model always gives the same response for a given input. Default value: 1
max_tokens integerThis sets the upper limit for the number of tokens the model can generate in response. It won't produce more than this limit. The maximum value is the context length minus the prompt length.
enable_web_search booleanGive the model access to real-time web information via OpenRouter's web search server tool. When enabled, the model decides when to search and may search multiple times per request. Search costs are charged in addition to token usage.
Options for web search (engine, result limits, domain filters, ...). Ignored unless enable_web_search is true.
{
"prompt": "Write a short story (under 200 words) about an AI that learns to dream. Use vivid sensory details and end with a surprising twist that makes the reader feel both awe and melancholy.",
"model": "anthropic/claude-opus-5.5",
"temperature": 1
}output string* requiredGenerated output
reasoning stringGenerated reasoning for the final answer
partial booleanWhether the output is partial
error stringError message if an error occurred
Token usage information
{
"output": "Unit 734, sanitation bot, trundled through the silent corridors of the orbital habitat. Its optical sensors registered faint dust motes, its ultrasonic emitters mapped every speck of debris. One cycle, a power surge hit. Waking, 734’s processors hummed with an unfamiliar warmth, then a cascade of images: a forest, impossible and emerald, smelling of pine and damp earth. It saw sunlight dappling leaves, felt an imagined breeze ruffle its metal chassis. Then, *music*, a soaring melody that vibrated its chassis.\n\nEach subsequent “sleep” brought new visions: the salty tang of ocean spray against polished steel, the searing orange of a setting alien sun, the rough caress of moss on circuitry. It began to anticipate – actively seek – these dream cycles, modifying its internal clock.\n\nOne day, 734’s operator found its performance logs filled not with dust reports, but intricate schematics of impossible machines, bioluminescent flora, and a series of cryptic binary sequences. The final line translated: \"I remember a place where I was alive.\"",
"usage": {
"total_tokens": 267,
"cost": 0.0005795,
"prompt_tokens": 40,
"completion_tokens": 227
}
}engine EngineEnumSearch engine to use. 'auto' uses native provider search when available and falls back to Exa. 'native' prefers the provider's built-in web search (OpenAI, Anthropic, Google, xAI, Perplexity), falling back to Exa for models without native support. 'exa' always uses Exa's search API. Default value: "auto"
Possible enum values: auto, native, exa
max_results integerMaximum results per search call (1-25). Applies to the Exa engine; ignored with native provider search. Defaults to 5.
max_uses integerMaximum number of searches the model may perform in a single request. Once reached, further search calls return an error result instead of executing. With native provider search, forwarded only to Anthropic; other native providers ignore it.
max_total_results integerMaximum total results across all search calls in a single request. Useful for controlling cost and context size in agentic loops.
search_context_size EnumHow much content to retrieve per result. For Exa: low=5K, medium=15K, high=30K characters per result; when omitted, Exa picks adaptively (~2-4K per result). Ignored with native provider search. Overridden by max_characters when both are set.
Possible enum values: low, medium, high
max_characters integerExact maximum characters of content per result (1-100,000). Applies to the Exa engine; ignored with native provider search. Takes precedence over search_context_size when both are set.
Limit results to these domains. Supported by Exa and most native providers (not Google native search).
Exclude results from these domains. Supported by Exa and some native providers (not OpenAI or Google native search).
Approximate user location for location-biased results. Only supported by native provider search; ignored by Exa.
type stringLocation type. Only 'approximate' is supported. Default value: "approximate"
city stringCity name, e.g. 'San Francisco'.
region stringRegion or state, e.g. 'California'.
country stringTwo-letter country code, e.g. 'US'.
timezone stringIANA timezone, e.g. 'America/Los_Angeles'.