avacache

Public, anonymous HTTPS archive of the Avalanche C-Chain as daily Parquet files.

Fetching manifest…

Every UTC day of the chain is served as three Parquet files — blocks, txs, and events — behind parquet.avacache.com. Fetch a full day, a range, or use HTTP range reads for column pushdown straight from the bucket.

Install

Pre-release: SDKs aren't on PyPI/npm yet — install from avasnap/avacache-sdk.

Python

pip install "git+https://github.com/avasnap/avacache-sdk.git#subdirectory=python"
from avacache import Client

with Client() as c:
    m = c.manifest()
    txs = c.load_day(m.latest_complete_date, "txs")

TypeScript — Node 18+ / browsers

git clone https://github.com/avasnap/avacache-sdk
npm install ./avacache-sdk/typescript
import { Client } from 'avacache';

const c = new Client();
const m = await c.manifest();
const txs = await c.loadDay(m.latest_complete_date, 'txs');

Dataset

Docs

DuckDB

Zero-install via httpfs:

INSTALL httpfs; LOAD httpfs;
SELECT COUNT(*) FROM read_parquet(
  'https://parquet.avacache.com/daily/2026-04-18.txs.parquet'
) WHERE status = 0;

Or via the SDK helper, which registers blocks, txs, and events as views:

from avacache.duckdb import open_day
con = open_day("2026-04-18")
con.execute("SELECT COUNT(*) FROM txs WHERE status = 0").fetchall()

Try it in your browser

No install, no server — this button fetches the events file for 2021-03-15 (1.6 MB, the early Pangolin era) directly from R2, decodes it with hyparquet, and ranks the top 10 pairs by WAVAX volume. Token symbols resolved live via Avalanche public RPC.