agentscope-ai/agentscope · error · RuntimeError

DashScope multimodal embedding API error: {res}

Error message

DashScope multimodal embedding API error: {res}

What it means

Raised when dashscope.MultiModalEmbedding.call returns a non-200 status code; the full result object is included in the message for root-causing (auth, quota, invalid media URL, model issues).

Source

Thrown at src/agentscope/embedding/_dashscope/_model.py:461

        if self.embedding_cache:
            cached = await self.embedding_cache.retrieve(
                identifier=cache_identifier,
            )
            if cached:
                return EmbeddingResponse(
                    embeddings=cached,
                    usage=EmbeddingUsage(tokens=0, time=0),
                    source="cache",
                )

        import dashscope

        start_time = datetime.now()
        res = dashscope.MultiModalEmbedding.call(**api_kwargs)
        time = (datetime.now() - start_time).total_seconds()

        if res.status_code != 200:
            raise RuntimeError(
                f"DashScope multimodal embedding API error: {res}",
            )

        embeddings = [entry["embedding"] for entry in res.output["embeddings"]]
        if self.embedding_cache:
            await self.embedding_cache.store(
                identifier=cache_identifier,
                embeddings=embeddings,
            )
        return EmbeddingResponse(
            embeddings=embeddings,
            usage=EmbeddingUsage(
                tokens=res.usage.get("image_tokens", 0)
                + res.usage.get("input_tokens", 0),
                time=time,
            ),
            source="api",
        )

View on GitHub (pinned to e90f1c7592)

Solutions

1. Inspect the embedded response for the status code and message, then address credentials, URL reachability, or size limits
2. Ensure media URLs are publicly reachable (or use base64 for images)
3. Retry transient 429/5xx with backoff and shrink the batch

Example fix

# before
res = await model(items)
# after
try:
    res = await model(items)
except RuntimeError as e:
    if "429" in str(e):
        await asyncio.sleep(5); res = await model(items)
    else:
        raise

urls = [b.source.url for b in blocks if hasattr(b.source, 'url')]
assert all(urllib.parse.urlparse(u).scheme in (\"http\", \"https\") for u in urls)

try:\n    await model(items)\nexcept RuntimeError as e:\n    if \"429\" in str(e) or \"timeout\" in str(e).lower(): await asyncio.sleep(2 ** n); await model(items)\n    else: raise

When it happens

Trigger: Calling multimodal embedding while DashScope rejects the request: bad API key, unreachable/expired media URLs, oversized payloads, rate limits, or an unavailable model.

Common situations: Embedding images referenced by internal URLs not reachable from DashScope; expired pre-signed URLs; large batches exceeding request limits; quota exhaustion.

Related errors

• DashScope text embedding API error: {response}
• Invalid input: {item!r}. Expected str or DataBlock.
• Multimodal embedding API only supports URL input for video d
• Unsupported media type {media_type!r} in DataBlock. Expected
• Text embedding model {self.model!r} only accepts str inputs,

AI-assisted analysis of agentscope-ai/agentscope@e90f1c7592 (2026-08-28). Data as JSON: /api/errors/6399df0dba70d4a5. Report an issue: GitHub.