LLM Aggregator API

FlowDot exposes an OpenAI-compatible HTTP API. Point any OpenAI SDK at https://flowdot.ai/api/v1 with a Bearer fd_agg_… key and you can call any supported model — OpenAI, Anthropic, Google, Redpill, OpenRouter, and more — through a single endpoint. It's the same idea as OpenRouter.

Quickstart

curl

curl https://flowdot.ai/api/v1/chat/completions \
  -H "Authorization: Bearer fd_agg_xxxxxxxxxxxxx" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "openai/gpt-4o-mini",
    "messages": [{"role": "user", "content": "Hello from FlowDot"}]
  }'

Python (openai SDK)

from openai import OpenAI

client = OpenAI(
    api_key="fd_agg_xxxxxxxxxxxxx",
    base_url="https://flowdot.ai/api/v1",
)

resp = client.chat.completions.create(
    model="anthropic/claude-sonnet-4-6",
    messages=[{"role": "user", "content": "Summarise FlowDot in one sentence."}],
)
print(resp.choices[0].message.content)

JavaScript / TypeScript (openai SDK)

import OpenAI from 'openai';

const client = new OpenAI({
  apiKey: process.env.FLOWDOT_AGG_KEY,
  baseURL: 'https://flowdot.ai/api/v1',
});

const resp = await client.chat.completions.create({
  model: 'redpill/google/gemini-2.5-flash-lite',
  messages: [{ role: 'user', content: 'Hello from FlowDot' }],
});
console.log(resp.choices[0].message.content);

List available models

curl https://flowdot.ai/api/v1/models \
  -H "Authorization: Bearer fd_agg_xxxxxxxxxxxxx"

Nested Model IDs

Model IDs are path-style, with the provider first. When an aggregator provider (like Redpill or OpenRouter) is the first segment, the rest of the path is forwarded upstream — hence the "double-nested" form.

Reasoning models (openai/o1, openai/o3, openai/gpt-5) accept max_completion_tokens rather than max_tokens and ignore temperature. The SDK handles this for you.

Call GET /api/v1/models for the authoritative live list of IDs you can use.

Three Funding Paths

You can pay for inference on FlowDot in three ways, independently:

All three can be mixed on one account — pick whichever makes sense per-model or per-integration.

Settings Modal — where each thing lives

Every user-level setting on FlowDot is managed from one place. Open any app page (Dashboard, Workflows, Agent, etc.) and click the settings gear icon in the top-right. The settings modal opens with these tabs:

Streaming

Pass "stream": true. The response is Server-Sent Events — data: {...}\n\n chunks terminated by data: [DONE], identical in shape to OpenAI's streaming response.

import OpenAI from 'openai';

const client = new OpenAI({
  apiKey: process.env.FLOWDOT_AGG_KEY,
  baseURL: 'https://flowdot.ai/api/v1',
});

const stream = await client.chat.completions.create({
  model: 'openai/gpt-4o-mini',
  messages: [{ role: 'user', content: 'Write a haiku about bees.' }],
  stream: true,
});

for await (const chunk of stream) {
  process.stdout.write(chunk.choices[0]?.delta?.content ?? '');
}

Tool Calling

The aggregator accepts OpenAI's tools / tool_choice shape and returns tool_calls on the assistant message. Passthrough behaviour depends on upstream support — OpenAI, Anthropic, and most Redpill / OpenRouter models all work; a handful of smaller open models do not.

const resp = await client.chat.completions.create({
  model: 'anthropic/claude-sonnet-4-6',
  messages: [{ role: 'user', content: 'What is the weather in Paris?' }],
  tools: [{
    type: 'function',
    function: {
      name: 'get_weather',
      description: 'Get current weather for a city',
      parameters: {
        type: 'object',
        properties: { city: { type: 'string' } },
        required: ['city'],
      },
    },
  }],
  tool_choice: 'auto',
});

console.log(resp.choices[0].message.tool_calls);

Errors

The aggregator returns OpenAI-shaped error objects. The ones specific to FlowDot:

All Aggregator Endpoints

All require Authorization: Bearer fd_agg_….

Dashboard-only key management (auth:sanctum, session)

These are not callable with an fd_agg_ token — they're for the web dashboard flow.