ModuleNotFoundError: No Module Named ‘duckdb’ (2026)

DuckDB is the embedded OLAP database that has replaced pandas + SQLite in many 2026 data pipelines. If you see ModuleNotFoundError: No module named ‘duckdb’, install with one pip command. No server required, no setup beyond pip.

Step 1: Install duckdb

pip install duckdb

# With extra extensions:
pip install 'duckdb[httpfs,parquet,json]'

# For Jupyter integration:
pip install duckdb jupysql duckdb-engine

Step 2: First query

import duckdb

# In-memory database, no file needed
result = duckdb.sql("SELECT 42 AS answer").fetchall()
print(result)  # [(42,)]

# Query a CSV directly without loading it
duckdb.sql("SELECT COUNT(*) FROM 'data.csv'").show()

Step 3: Query Parquet on S3 (zero copy)

import duckdb

con = duckdb.connect()
con.sql("INSTALL httpfs; LOAD httpfs;")
con.sql("""
    SELECT COUNT(*) FROM read_parquet('s3://my-bucket/2026/*.parquet')
    WHERE event_type = 'purchase'
""").show()

Step 4: Polars and pandas interop

import pandas as pd
import polars as pl
import duckdb

# Query a pandas DataFrame directly by variable name
df = pd.DataFrame({'a': [1, 2, 3]})
duckdb.sql("SELECT * FROM df WHERE a > 1").show()

# Same with polars
pl_df = pl.DataFrame({'a': [1, 2, 3]})
duckdb.sql("SELECT * FROM pl_df WHERE a > 1").show()

# Return as polars
duckdb.sql("SELECT 1 AS x").pl()

Why this error happens