Mintplex-Labs/anything-llm · error · Error

GroqAI:chatCompletion

Error message

GroqAI:chatCompletion: ${this.model} is not valid for chat completion!

What it means

Guard at the top of GroqLLM.getChatCompletion(): it awaits isValidChatCompletionModel(this.model) and throws if that returns false. In this codebase isValidChatCompletionModel is simply `return !!modelName` (server/utils/AiProviders/groq/index.js:60), so it validates only that a model name exists — it does NOT check the model against Groq's live model list. Because the constructor falls back `modelPreference || GROQ_MODEL_PREF || "llama-3.1-8b-instant"`, this.model is almost never falsy, making this a defensive near-unreachable check.

Solutions

Inspect what this.model actually resolved to (constructor default is llama-3.1-8b-instant) — if you truly got this throw, it is empty, so set GROQ_MODEL_PREF or pass a modelPreference when constructing.
Set GROQ_MODEL_PREF in the backend .env to a currently listed Groq model id (check console.groq.com/models).
If the intent was 'model rejected by Groq', note this guard cannot produce that; the actual API rejection is rethrown from the .catch inside getChatCompletion with the SDK's message — debug that instead.
Upgrade AnythingLLM: newer Groq models rotate frequently; decommissioned ids fail at request time, so pin a supported id.

Example fix

// before
const llm = new GroqLLM(embedder, ""); // model falls back to env/default

// after — pin an explicit, current model
const llm = new GroqLLM(embedder, "llama-3.3-70b-versatile");

Defensive patterns

Strategy: validation

Validate before calling

const model = modelPreference || process.env.GROQ_MODEL_PREF || "llama-3.1-8b-instant";
if (!model) throw new Error("No Groq model resolved");
if (!(await llm.isValidChatCompletionModel(model))) throw new Error(`Model ${model} rejected`);
const res = await llm.getChatCompletion(messages);

Type guard

/** True when the instance can safely serve non-stream completions. */
function isRunnableGroqInstance(llm) {
  return llm instanceof GroqLLM && typeof llm.model === "string" && llm.model.length > 0;
}

Try / catch

try {
  await llm.getChatCompletion(messages);
} catch (e) {
  if (/not valid for chat completion/i.test(e.message)) {
    // model state lost — rebuild instance with an explicit model
  } else throw e;
}

Prevention

Always pass a non-empty modelPreference or set GROQ_MODEL_PREF so the constructor default chain is never relied on.
Treat this message as a state-corruption signal: log llm.model whenever it fires.
Pin model ids you have verified against console.groq.com/models; Groq decommissions ids frequently.

When it happens

Trigger: Calling llm.getChatCompletion(messages) on a GroqLLM instance whose this.model is an empty string or null at call time. Practically reachable only if the instance was constructed with an explicitly falsy-but-present model and the env default chain was bypassed (e.g., a mutated instance or a monkey-patched constructor default).

Common situations: Almost never seen in production because the constructor's || chain guarantees a default model name. Developers encounter the string in logs usually as a red herring when the real failure is the Groq API rejecting the model id server-side (404 model_decommissioned), which surfaces later, not from this guard.

Related errors

AI-assisted analysis of Mintplex-Labs/anything-llm@f92433b4ea (2026-08-18). Data as JSON: /api/errors/8e8822ec14c7245b. Report an issue: GitHub.

Appendix: source

Thrown at server/utils/AiProviders/groq/index.js:171

    systemPrompt = "",
    contextTexts = [],
    chatHistory = [],
    userPrompt = "",
    attachments = [], // This is the specific attachment for only this prompt
  }) {
    // NOTICE: SEE GroqLLM.#conditionalPromptStruct for more information on how attachments are handled with Groq.
    return this.#conditionalPromptStruct({
      systemPrompt,
      contextTexts,
      chatHistory,
      userPrompt,
      attachments,
    });
  }

  async getChatCompletion(messages = null, { temperature = 0.7 }) {
    if (!(await this.isValidChatCompletionModel(this.model)))
      throw new Error(
        `GroqAI:chatCompletion: ${this.model} is not valid for chat completion!`
      );

    const result = await LLMPerformanceMonitor.measureAsyncFunction(
      this.openai.chat.completions
        .create({
          model: this.model,
          messages,
          temperature,
        })
        .catch((e) => {
          throw new Error(e.message);
        })
    );

    if (
      !result.output.hasOwnProperty("choices") ||
      result.output.choices.length === 0

View on GitHub (pinned to f92433b4ea)