What causes "xAI API key is required" in mem0ai/mem0?

Export XAI_API_KEY in the environment where Node runs: export XAI_API_KEY=... (or add it to your deployment's env config).

mem0ai/mem0 · error

xAI API key is required

Error message

xAI API key is required

What it means

Thrown by the XAILLM constructor when neither config.apiKey nor the XAI_API_KEY environment variable is present. xAI's Grok API requires authentication, so the constructor fails fast instead of letting every request fail later with a 401. This mirrors the Python SDK's mem0/llms/xai.py behavior.

Source

Thrown at mem0-ts/src/oss/src/llms/xai.ts:18

import { OpenAILLM } from "./openai";
import { LLMConfig, Message } from "../types";
import { LLMResponse } from "./base";

/**
 * xAI (Grok) LLM provider.
 *
 * xAI's Grok API is OpenAI-compatible, so this simply reuses {@link OpenAILLM}
 * and overrides the connection defaults — mirroring `mem0/llms/xai.py` in the
 * Python SDK. The API key resolves from `config.apiKey` or the `XAI_API_KEY`
 * env var, and the base URL from `config.baseURL`, `XAI_API_BASE`, else
 * `https://api.x.ai/v1`.
 */
export class XAILLM extends OpenAILLM {
  constructor(config: LLMConfig) {
    const apiKey = config.apiKey || process.env.XAI_API_KEY;
    if (!apiKey) {
      throw new Error("xAI API key is required");
    }
    super({
      ...config,
      apiKey,
      baseURL:
        config.baseURL || process.env.XAI_API_BASE || "https://api.x.ai/v1",
      model: config.model || "grok-4.3",
    });
  }

  async generateResponse(
    messages: Message[],
    responseFormat?: { type: string },
    tools?: any[],
  ): Promise<string | LLMResponse> {
    try {
      return await super.generateResponse(messages, responseFormat, tools);
    } catch (err) {

View on GitHub (pinned to 001c235229)

Solutions

Export XAI_API_KEY in the environment where Node runs: export XAI_API_KEY=... (or add it to your deployment's env config).
Or pass the key explicitly: new Memory({ llm: { provider: 'xai', config: { apiKey: process.env.XAI_API_KEY } } }).
If using .env, ensure dotenv (or the runtime's env loading) runs before the Memory/XAILLM constructor executes.
Verify with a quick check before constructing: if (!process.env.XAI_API_KEY) throw new Error('XAI_API_KEY missing');

Example fix

// before
const memory = new Memory({
  llm: { provider: 'xai', config: {} }, // throws: xAI API key is required
});

// after
const memory = new Memory({
  llm: {
    provider: 'xai',
    config: { apiKey: process.env.XAI_API_KEY }, // or export XAI_API_KEY
  },
});

Defensive patterns

Strategy: validation

Validate before calling

const xaiKey = process.env.XAI_API_KEY;
if (!xaiKey) {
  throw new Error('XAI_API_KEY is not set — add it to the environment before constructing Memory');
}

Type guard

function hasXaiCredentials(config: { apiKey?: string } | undefined): boolean {
  return Boolean(config?.apiKey || process.env.XAI_API_KEY);
}

Try / catch

try {
  memory = new Memory({ llm: { provider: 'xai', config: { apiKey: process.env.XAI_API_KEY } } });
} catch (err) {
  if (err instanceof Error && err.message === 'xAI API key is required') {
    // fail startup with an operator-actionable message
    process.exitCode = 1;
    throw new Error('Missing XAI_API_KEY: set it in the deployment environment');
  }
  throw err;
}

Prevention

Declare XAI_API_KEY in Docker/CI/serverless env specs so it is never silently absent.
Assert required env vars at process start with a single config-validation module.
Add a startup self-test that constructs Memory once, so key errors surface at boot, not mid-request.

When it happens

Trigger: Constructing Memory with llm.provider 'xai' (or instantiating XAILLM directly) without passing apiKey in config.llm.config and without XAI_API_KEY exported in the process environment. Throws synchronously at construction time, before any memory operation runs.

Common situations: Env var set in a shell but not in the deployment (Docker/CI/serverless where env must be declared), .env file present but not loaded before import, typo in the variable name (XAI_KEY, X_AI_API_KEY), or assuming the SDK reads a differently-named variable.

Related errors

AI-assisted analysis of mem0ai/mem0@001c235229 (2026-08-15). Data as JSON: /api/errors/0e34ba269c625455. Report an issue: GitHub.