Chat with PDF API in India

Ask questions about your PDFs

Manually searching through PDFs is tedious. unoblox makes it easy to build conversational PDF interfaces—letting users ask questions and get instant answers—via one OpenAI-compatible endpoint, billed in rupees.

Why PDF chat?

PDFs contain knowledge (reports, contracts, manuals) but are hard to search. AI-powered chat lets you extract, summarize, and answer questions about document content in natural language.

Best models for document Q&A

Model	Input ₹	Output ₹	Strength
Claude Opus	₹504	₹2520	Long context, precise extraction
GPT-4.1	₹201.6	₹806.4	Fast, good reasoning
Qwen3 Max	₹120.95	₹604.77	Cost-effective, large context

Prices per 1M tokens. Long documents need high context windows.

Implementation workflow

from openai import OpenAI
import PyPDF2

client = OpenAI(
    api_key="ub-gw-...",
    base_url="https://api.unoblox.ai/v1"
)

# Extract text from PDF
def extract_pdf_text(pdf_path):
    text = ""
    with open(pdf_path, "rb") as f:
        reader = PyPDF2.PdfReader(f)
        for page in reader.pages:
            text += page.extract_text()
    return text

pdf_text = extract_pdf_text("document.pdf")
user_question = "What is the main conclusion?"

response = client.chat.completions.create(
    model="anthropic/claude-opus-4-5",
    messages=[
        {"role": "system", "content": f"Document:\n{pdf_text}"},
        {"role": "user", "content": user_question}
    ]
)

print(response.choices[0].message.content)

Common use cases

• Financial report analysis (investor Q&A)
• Legal contract review and Q&A
• Academic paper search and summarization
• Customer documentation support
• Insurance policy lookup
• Product manual search

Frequently asked questions

Q: What's the PDF size limit? No hard limit, but model context windows cap it. Claude Opus supports 200K tokens (~150 pages of text). Split large PDFs into sections if needed.

Q: Can I index multiple PDFs? Yes. Extract text from all PDFs, concatenate into sections, and include a table of contents so the model can reference specific documents.

Q: How do I handle scanned/image PDFs? Use GPT-4o with vision to extract text from scanned images first, then feed to your Q&A model. Cost is ~₹1-2 per page.

Q: Can I cite the source (page/section)? Yes. Include page numbers in the extracted text. Ask the model to cite sources in its answer.

Q: What about proprietary/sensitive documents? All data stays within India (data residency). Logs retained per your SLA (default 90 days). No third-party access. Review our security policy.

Q: How much does it cost per document? At Claude Opus rates, a 50-page document (~25K tokens) + one question costs ~₹12.6 (input) + ₹2.5 (output). Reuse the document context across many questions to amortize cost.

Get started in rupees → https://unoblox.ai/sign-in