The same report, represented as content the model can read.
Original PDF Page 2
LlamaParse agentic
Quarterly operating results
CHF thousands / Revenue and operating costs by site
Site
Q1 2026
Q2 2026
Revenue
Costs
Revenue
Costs
Lausanne
120
80
150
90
Sion
95
60
140
70
Neuchatel
80
55
100
65
Total
295
195
390
225
Operating costs exclude financing costs and corporate income tax.
Actual output recorded on 11 Nov 2026 · Illustrative report · Tables / Charts ↗
LlamaParse / Python
Upload. Parse. Read the output.
llama-cloudPython
from llama_cloud import LlamaCloud
with LlamaCloud() as client:
file = client.files.create(
file="data/alpine_solar.pdf", purpose="parse",
)
result = client.parsing.parse(
file_id=file.id, tier="agentic",
version="latest", expand=["markdown"],
)
for page in result.markdown.pages:
print(page.page_number, page.markdown)
Readable text can still lose relationships that matter to an answer.
Original PDF Page 2
LiteParse Markdown + OCR
Quarterly operating results
CHF thousands / Revenue and operating costs by site
Q1 2026 Q2 2026 Site
Revenue Costs Revenue Costs
Lausanne
120
80
150
90
Sion
95
60
140
70
Neuchatel
80
55
100
65
Total
295
195
390
225
Operating costs exclude financing costs and corporate income tax.
Actual output recorded on 11 Nov 2026 · This notebook sends parsed text, not page images, to the answering model.
LiteParse / Python
A local parse in a few lines
Same PDF. A different way to turn it into text.
liteparsePython
from liteparse import LiteParse
parser = LiteParse(output_format="markdown", ocr_enabled=True)
result = parser.parse("data/alpine_solar.pdf")
for page in result.pages:
print(page.page_num, page.markdown or page.text)
Look at the parsed content before building the index.