Mistral

OverviewCopy Anchor Link

Mistral AI's chat models.

• MistralLLMService: an OpenAILLMService subclass with Mistral's base URL and default model. It uses MistralLLMAdapter, which rewrites the history to Mistral's rules: an assistant message after every tool message, system messages only at the start, and prefix: true on a trailing assistant message.

InstallationCopy Anchor Link

bun add @pipedog/mistral

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

import { MistralLLMService } from "@pipedog/mistral/llm";
import { MistralLLMAdapter } from "@pipedog/mistral/mistral_adapter";

PrerequisitesCopy Anchor Link

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

MistralLLMServiceCopy 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 MistralLLMService.Settings({...}) (MistralLLMSettings). 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 { MistralLLMService } from "@pipedog/mistral/llm";

const llm = new MistralLLMService({
  apiKey: process.env.MISTRAL_API_KEY ?? "",
  settings: new MistralLLMService.Settings({
    model: "mistral-small-latest",
    systemInstruction: "You are a helpful assistant in a voice conversation.",
    temperature: 0.7,
  }),
});

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 MistralLLMService.Settings({ temperature: 0.2, maxTokens: 300 }),
});

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).
• Messages go through MistralLLMAdapter, which rewrites the history to Mistral's rules: an assistant message after every tool message, system messages only at the start, and prefix: true on a trailing assistant message.
• seed is sent as Mistral's random_seed. stream_options, max_completion_tokens and service_tier are not sent.
• Mistral detects function calls from the whole history, so calls that already have a tool result in the context are skipped instead of run twice. on_function_calls_started may still fire twice.
• The same package has Mistral's speech-to-text and text-to-speech services.

Runtime notesCopy Anchor Link

Generated from pipecat.runtimes and pipecat.compat in packages/mistral/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 Mistral itself.