Mintplex-Labs/anything-llm · error · Error

LMStudio Failed to embed: ${Array.from(uniqueErrors).join(",

Error message

LMStudio Failed to embed: ${Array.from(uniqueErrors).join(", ")}

What it means

Thrown at the end of embedChunks (line 110) when at least one per-chunk embedding request failed. Because LMStudio drops concurrent requests, embedChunks processes sequentially; on the first error hasError is set and the loop breaks. Errors are collected, deduplicated into [type]: message strings, and the whole batch is aborted since partial data would be incomplete. error.type comes from response.data.error.code, HTTP status, or 'failed_to_embed'; a missing embedding array yields type 'EMPTY_ARR'.

Source

Thrown at server/utils/EmbeddingEngines/lmstudio/index.js:110

      );
    }

    // Accumulate errors from embedding.
    // If any are present throw an abort error.
    const errors = results
      .filter((res) => !!res.error)
      .map((res) => res.error)
      .flat();

    if (errors.length > 0) {
      let uniqueErrors = new Set();
      console.log(errors);
      errors.map((error) =>
        uniqueErrors.add(`[${error.type}]: ${error.message}`)
      );

      if (errors.length > 0)
        throw new Error(
          `LMStudio Failed to embed: ${Array.from(uniqueErrors).join(", ")}`
        );
    }

    const data = results.map((res) => res?.data || []);
    return data.length > 0 ? data : null;
  }
}

module.exports = {
  LMStudioEmbedder,
};

View on GitHub (pinned to 526360e320)

Solutions

1. Match the [type] prefix: EMPTY_ARR means the model returned no vector (verify the loaded model is an embedding model, not a chat model); an HTTP status means check LMStudio logs for that request
2. curl the LMStudio /v1/embeddings endpoint with the exact model and a sample chunk to reproduce
3. Lower the document chunk size so no chunk exceeds the embedding model's max context
4. Ensure no other process is restarting/unloading the LMStudio model mid-batch

Example fix

// before
EMBEDDING_MODEL_PREF=some-chat-model   // not an embedding model -> EMPTY_ARR

// after
EMBEDDING_MODEL_PREF=nomic-ai/nomic-embed-text-v1.5

// confirm the model returns vectors before the bulk run
async function lmstudioEmbeds(openai, model, sample = 'hello') {
  const r = await openai.embeddings.create({ model, input: sample, encoding_format: 'base64' });
  return Array.isArray(r.data?.[0]?.embedding) && r.data[0].embedding.length > 0;
}

function isLMStudioEmbedError(e) {
  return e instanceof Error && /LMStudio Failed to embed/.test(e.message);
}

try {
  return await embedder.embedChunks(chunks);
} catch (e) {
  if (/EMPTY_ARR/.test(e.message)) {
    throw new Error('Loaded LMStudio model is not an embedding model', { cause: e });
  }
  throw e;
}

When it happens

Trigger: A sequential embeddings.create call rejecting: LMStudio returns 404/500 for the model name; the response.data[0].embedding is missing/empty (throws {type:'EMPTY_ARR'} inline at line 76); context length exceeded for the chunk; LMStudio crashed mid-batch; encoding_format base64 not supported by the loaded model.

Common situations: EMBEDDING_MODEL_PREF does not match the loaded model identifier; chunk too long for the embedding model's context; LMStudio OOM or model swap mid-run; LMStudio version that mishandles base64 encoding_format; concurrent AnythingLLM jobs hitting the single-threaded LMStudio server.

Related errors

• LiteLLM Failed to embed: ${error}
• LMStudio service could not be reached. Is LMStudio running?
• LocalAI Failed to embed: ${error}
• Ollama Failed to embed: ${error}
• OpenAI Failed to embed: ${error}

AI-assisted analysis of Mintplex-Labs/anything-llm@526360e320 (2026-08-13). Data as JSON: /api/errors/88ecd3f8bf9e205b. Report an issue: GitHub.