Inception

OverviewCopy Anchor Link

Inception serves Mercury-3, a diffusion-based reasoning language model, over an OpenAI-compatible API.

• InceptionLLMService: an OpenAILLMService subclass with Inception's base URL and default model.

InstallationCopy Anchor Link

bun add @pipedog/inception

It depends on @pipedog/openai, which is installed with it.

import { InceptionLLMService } from "@pipedog/inception/llm";

PrerequisitesCopy Anchor Link

• An Inception account and an API key from Inception.
• Pass the key as apiKey. It is required, and the service never falls back to OPENAI_API_KEY. The examples read it from INCEPTION_API_KEY.

InceptionLLMServiceCopy Anchor Link

ConfigurationCopy Anchor Link

The other options (organization, project, defaultHeaders, serviceTier, retryOnTimeout, retryTimeoutSecs, function calling) are OpenAILLMService's. See OpenAI.

SettingsCopy Anchor Link

Runtime-updatable settings, passed as settings: new InceptionLLMService.Settings({...}) (InceptionLLMSettings). They are OpenAI's settings, with these defaults:

Every other setting keeps OpenAI's default. Unset sampling settings are left out of the request, so the provider's own defaults apply. extra carries provider-specific request fields.

UsageCopy Anchor Link

import { LLMUpdateSettingsFrame } from "@pipedog/core/frames/frames";
import { InceptionLLMService } from "@pipedog/inception/llm";

const llm = new InceptionLLMService({
  apiKey: process.env.INCEPTION_API_KEY ?? "",
  settings: new InceptionLLMService.Settings({
    model: "mercury-3",
    systemInstruction: "You are a helpful assistant in a voice conversation.",
  }),
});

Trade reasoning for speed with reasoningEffort:

const fast = new InceptionLLMService({
  apiKey: process.env.INCEPTION_API_KEY ?? "",
  settings: new InceptionLLMService.Settings({ reasoningEffort: "instant" }),
});

Change settings mid-call with an LLMUpdateSettingsFrame. The next completion uses them:

// Queue it on the worker, or push it from a processor upstream of the service.
const update = new LLMUpdateSettingsFrame({
  delta: new InceptionLLMService.Settings({ reasoningEffort: "low" }),
});

NotesCopy Anchor Link

• Runs a completion for each LLMContextFrame, between an LLMFullResponseStartFrame and an LLMFullResponseEndFrame, and streams the reply as LLMTextFrames, as OpenAILLMService does. Function calls, token usage metrics and runInference work the same way.
• "developer" messages are sent as "user" messages, since the API doesn't support that role (supportsDeveloperRole is false).
• realtime and reasoningEffort go out in the request's extra_body. A field you put in settings.extra under extra_body wins over the service's own.
• The default model was mercury-2 before. Inception retires Mercury 2 and 2.5 on November 15, 2026, and they are closed to new signups. Until then, settings: new InceptionLLMService.Settings({ model: "mercury-2" }) keeps it.

Runtime notesCopy Anchor Link

Generated from pipecat.runtimes and pipecat.compat in packages/inception/package.json: ✅ means the compat smoke test passes on that runtime.

• The client is the official openai JS SDK (through @pipedog/openai), which talks through fetch. It works on Bun, Node, Deno and Cloudflare Workers.
• The deprecated model constructor option is not ported. Use settings.
• Tested against a local OpenAI-compatible mock server, not against Inception itself.