Mintplex-Labs/anything-llm · error

Ollama returned empty embeddings for batch!

Error message

Ollama returned empty embeddings for batch!

What it means

Ollama answered the /api/embed call with HTTP 200, but the response's `embeddings` field was missing or an empty array. Because the code sends the batch via the `input` parameter, it expects exactly one embedding per input string; zero results means the model ran but produced nothing. The most common cause is that the configured model is a generative/LLM model with no embedding capability, or the Ollama version is too old to return the batch `embeddings` shape.

Solutions

1. Point EMBEDDING_MODEL_PREF at a real embedding model (nomic-embed-text, mxbai-embed-large, all-minilm, snowflake-arctic-embed) and `ollama pull` it.
2. Match the model name from `ollama list` exactly, including the tag.
3. Upgrade Ollama to a recent release so /api/embed with `input` arrays returns `embeddings`.
4. Reproduce outside AnythingLLM: `curl http://localhost:11434/api/embed -d '{"model":"nomic-embed-text","input":["hi"]}'` and confirm embeddings is non-empty.

Example fix

# .env — before (llama3 is a chat model and cannot embed)
EMBEDDING_MODEL_PREF=llama3

# .env — after
EMBEDDING_MODEL_PREF=nomic-embed-text

async function modelProducesEmbeddings(basePath, model) {
  const res = await fetch(`${basePath}/api/embed`, {
    method: "POST",
    headers: { "Content-Type": "application/json" },
    body: JSON.stringify({ model, input: ["ping"] }),
  });
  const body = await res.json();
  return Array.isArray(body.embeddings) && body.embeddings.length > 0;
}
if (!(await modelProducesEmbeddings(basePath, model))) {
  throw new Error(`${model} is not an embedding model — pick one from \`ollama list\` that embeds.`);
}

function hasEmbeddings(payload) {
  return (
    payload !== null &&
    typeof payload === "object" &&
    Array.isArray(payload.embeddings) &&
    payload.embeddings.length > 0 &&
    payload.embeddings.every((e) => Array.isArray(e))
  );
}

try {
  const vectors = await embedder.embedChunks(chunks);
} catch (err) {
  if (err.message.includes("empty embeddings")) {
    // model is not an embedding model: fix EMBEDDING_MODEL_PREF, do not retry
  } else {
    throw err;
  }
}

When it happens

Trigger: EMBEDDING_MODEL_PREF points at a chat model (llama3, mistral, qwen) instead of an embedding model; a model-name/tag mismatch so Ollama resolves a different model; an Ollama version that predates the /api/embed batch endpoint's response shape; a server returning an error-shaped body with status 200.

Common situations: Reusing the same model name for the LLM and the embedder in settings; forgetting `ollama pull nomic-embed-text` on a new machine; very old Ollama installs on NAS devices or routers.

Understand the failure class

Background: "empty response", "returned no data", "empty embeddings": what HTTP 200-with-empty-body errors mean across libraries — this error's family across 36 libraries.

Related errors

• Mistral returned empty embeddings for batch
• No embedding model was set.
• Ollama Failed to embed
• Ollama service could not be reached. Is Ollama running?
• ChromaCloud::Embedding dimension too large

AI-assisted analysis of Mintplex-Labs/anything-llm@3aec848f28 (2026-08-18). Data as JSON: /api/errors/c4fefaea93306e88. Report an issue: GitHub.

Appendix: source

Thrown at server/utils/EmbeddingEngines/ollama/index.js:115

    for (let i = 0; i < textChunks.length; i += this.maxConcurrentChunks) {
      const batch = textChunks.slice(i, i + this.maxConcurrentChunks);
      currentBatch++;

      try {
        // Use input param instead of prompt param to support batch processing
        const res = await this.client.embed({
          model: this.model,
          input: batch,
          options: {
            // Always set the num_ctx to the max chunk length defined by the user in the settings
            // so that the maximum context window is used and content is not truncated.
            num_ctx: this.embeddingMaxChunkLength,
          },
        });

        const { embeddings } = res;
        if (!Array.isArray(embeddings) || embeddings.length === 0)
          throw new Error("Ollama returned empty embeddings for batch!");

        // Using prompt param in embed() would return a single embedding (number[])
        // but input param returns an array of embeddings (number[][]) for batch processing.
        // This is why we spread the embeddings array into the data array.
        data.push(...embeddings);
        reportEmbeddingProgress(data.length, textChunks.length);
        this.log(
          `Batch ${currentBatch}/${totalBatches}: Embedded ${embeddings.length} chunks. Total: ${data.length}/${textChunks.length}`
        );
      } catch (err) {
        this.log(err.message);
        error = err.message;
        data = [];
        break;
      }
    }

    if (!!error) throw new Error(`Ollama Failed to embed: ${error}`);

View on GitHub (pinned to 3aec848f28)