When using larger pdfs (e.g. 20MB) from AWS S3 as signed urls in the parts then I receive the error:
ApiError: {“error”:{“code”:400,“message”:“The document has no pages.”,“status”:“INVALID_ARGUMENT”}}
The api call looks like this:
vertexAI.models.generateContent({
model: 'gemini-3-flash-preview',
contents: [
{
role: 'user',
parts: [
{"fileData":{"fileUri":"https://abc.s3.eu-central-1.amazonaws.com/abc/dcg.pdf?X-Amz-Algorithm=AWS4-HMAC-SHA256&X-Amz-Content-Sha256=UNSIGNED-PAYLOAD&X-Amz-Credential=abc%2F20260116%2Feu-central-1%2Fs3%2Faws4_request&X-Amz-Date=20260116T151901Z&X-Amz-Expires=600&X-Amz-Signature=abc&X-Amz-SignedHeaders=host&x-amz-checksum-mode=ENABLED&x-id=GetObject","mimeType":"application/pdf"}}, { text: "user-prompt" }],
},
],
config: {
responseMimeType: 'application/json',
responseSchema,
seed: this.SEED,
systemInstruction: [{ text: "systemPrompt }],
thinkingConfig: {
thinkingLevel: ThinkingLevel.HIGH,
},
},
})
I’m using version 1.37.0 of genai/js
In VertexAI Studio it’s the same!
Also when using VertexAI Studio and entering the file via the link
I get the same error
Opening the link in the browser works fine and the pdf looks good, also in Adobe