Mintplex-Labs/anything-llm · error

No Lemonade Model Pref was set.

Error message

No Lemonade Model Pref was set.

What it means

LemonadeLLM requires a model id before construction completes: it throws when neither the modelPreference argument nor LEMONADE_LLM_MODEL_PREF is present. The check runs before the OpenAI client is even built, and this.model = modelPreference || LEMONADE_LLM_MODEL_PREF afterwards. Lemonade needs a concrete model tag (e.g., a GGUF/Hybrid variant name it has loaded) for every completion request.

Solutions

1. Set LEMONADE_LLM_MODEL_PREF to the exact model tag Lemonade has loaded (query its models endpoint or server logs).
2. Or pass the model explicitly: new LemonadeLLM(embedder, "Llama-3.2-1B-Instruct-Hybrid").
3. Confirm spelling/case matches the server's registered tag, then restart AnythingLLM.

Example fix

# before (.env)
LEMONADE_LLM_BASE_PATH=http://127.0.0.1:8000

# after (.env)
LEMONADE_LLM_BASE_PATH=http://127.0.0.1:8000
LEMONADE_LLM_MODEL_PREF=Llama-3.2-1B-Instruct-Hybrid

const lemonadeModel = modelPreference || process.env.LEMONADE_LLM_MODEL_PREF;
if (!lemonadeModel) throw new Error("Set LEMONADE_LLM_MODEL_PREF or pass modelPreference");
const llm = new LemonadeLLM(embedder, lemonadeModel);

function hasLemonadeModel(arg) {
  return typeof (arg || process.env.LEMONADE_LLM_MODEL_PREF) === "string" && (arg || process.env.LEMONADE_LLM_MODEL_PREF).length > 0;
}

try {
  new LemonadeLLM(embedder, pref);
} catch (e) {
  if (/Lemonade Model Pref/i.test(e.message)) throw new Error("LEMONADE_LLM_MODEL_PREF unset — add the loaded model tag");
  throw e;
}

When it happens

Trigger: new LemonadeLLM(embedder) with no second argument while LEMONADE_LLM_MODEL_PREF is unset — i.e., choosing the Lemonade provider with only LEMONADE_LLM_BASE_PATH configured. The guard is `!env && !arg`, so unlike KoboldCPP's ?? variant, an empty-string modelPreference here still falls through to env; the throw needs both to be absent.

Common situations: Env templates that ship the base path but comment out the model line; using a model name from a different Lemonade install/version (wrong, but that fails later at HTTP, not here); programmatic construction that assumes a default model exists — Lemonade has none.

Related errors

• KoboldCPP must have a valid model set.
• LiteLLM must have a valid model set.
• LMStudio must have a valid model set.
• No Lemonade API Base Path was set.
• KoboldCPP must have a valid base path to use for the api.

AI-assisted analysis of Mintplex-Labs/anything-llm@3aec848f28 (2026-08-18). Data as JSON: /api/errors/3bcfb50bc62c06fa. Report an issue: GitHub.

Appendix: source

Thrown at server/utils/AiProviders/lemonade/index.js:17

const { NativeEmbedder } = require("../../EmbeddingEngines/native");
const {
  handleDefaultStreamResponseV2,
  formatChatHistory,
} = require("../../helpers/chat/responses");
const {
  LLMPerformanceMonitor,
} = require("../../helpers/chat/LLMPerformanceMonitor");
const { OpenAI: OpenAIApi } = require("openai");
const { humanFileSize } = require("../../helpers");

class LemonadeLLM {
  constructor(embedder = null, modelPreference = null) {
    if (!process.env.LEMONADE_LLM_BASE_PATH)
      throw new Error("No Lemonade API Base Path was set.");
    if (!process.env.LEMONADE_LLM_MODEL_PREF && !modelPreference)
      throw new Error("No Lemonade Model Pref was set.");

    this.className = "LemonadeLLM";
    this.lemonade = new OpenAIApi({
      baseURL: parseLemonadeServerEndpoint(
        process.env.LEMONADE_LLM_BASE_PATH,
        "openai"
      ),
      apiKey: process.env.LEMONADE_LLM_API_KEY || null,
    });

    this.model = modelPreference || process.env.LEMONADE_LLM_MODEL_PREF;
    this.embedder = embedder ?? new NativeEmbedder();
    this.defaultTemp = 0.7;

    // We can establish here since we cannot dynamically curl the context window limit from the API.
    this.limits = {
      history: this.promptWindowLimit() * 0.15,
      system: this.promptWindowLimit() * 0.15,

View on GitHub (pinned to 3aec848f28)