Full Example

#!/usr/bin/env python3
"""Read the current GPU price index, pull history, and overlay it on the
long-run informational baseline."""

import os
import requests
import pandas as pd

BASE = "https://data-api.compute-index.com/v1"
TOKEN = os.environ["TOKEN"]
headers = {"Authorization": f"Bearer {TOKEN}"}

GPU_FAMILY = "hopper"   # or "blackwell"
REGION = "US"

# Step 1: Latest blended transaction index (the settlement-benchmark series)
resp = requests.get(
    f"{BASE}/txn-index/current",
    headers=headers,
    params={
        "gpu_family": GPU_FAMILY,
        "region_group": REGION,
        "contract_type": "combined",
        "index_type": "blended_sigmoid",
    },
    timeout=30,
)
resp.raise_for_status()
latest = resp.json()["data"][0]
print(f"{GPU_FAMILY}-{REGION}: ${latest['index_value']:.4f}/GPU/hr on {latest['date']}")

# Step 2: Full transaction history (up to the 730-day window)
resp = requests.get(
    f"{BASE}/txn-index/history",
    headers=headers,
    params={
        "gpu_family": GPU_FAMILY,
        "region_group": REGION,
        "contract_type": "combined",
        "index_type": "blended_sigmoid",
        "days_back": 730,
    },
    timeout=30,
)
resp.raise_for_status()
txn = pd.DataFrame(resp.json()["data"])
txn["date"] = pd.to_datetime(txn["date"])

# Step 3: Extend history with the informational baseline (no days_back cap,
# real data back to early 2024)
resp = requests.get(
    f"{BASE}/info-index/history",
    headers=headers,
    params={"index_name": GPU_FAMILY, "region_group": REGION},
    timeout=30,
)
resp.raise_for_status()
info = pd.DataFrame(resp.json()["data"])
info["date"] = pd.to_datetime(info["date"])

# Step 4: Merge into one frame — informational baseline + transaction overlay
merged = (
    info[["date", "index_value"]]
    .rename(columns={"index_value": "informational"})
    .merge(
        txn[["date", "index_value"]].rename(columns={"index_value": "transaction"}),
        on="date",
        how="left",
    )
    .sort_values("date")
)

print(f"\nInformational baseline: {len(info)} days "
      f"({info['date'].min().date()} → {info['date'].max().date()})")
print(f"Transaction overlay:    {len(txn)} days "
      f"({txn['date'].min().date()} → {txn['date'].max().date()})")
print(merged.tail())

# merged.plot(x="date", y=["informational", "transaction"])  # with matplotlib

# Step 1: latest blended transaction index (Hopper-US, combined)
curl -H "Authorization: Bearer $TOKEN" \
  "https://data-api.compute-index.com/v1/txn-index/current?gpu_family=hopper&region_group=US&contract_type=combined&index_type=blended_sigmoid"

# Step 2: full transaction history (up to 730 days)
curl -H "Authorization: Bearer $TOKEN" \
  "https://data-api.compute-index.com/v1/txn-index/history?gpu_family=hopper&region_group=US&contract_type=combined&index_type=blended_sigmoid&days_back=730"

# Step 3: long-run informational baseline (no days_back cap)
curl -H "Authorization: Bearer $TOKEN" \
  "https://data-api.compute-index.com/v1/info-index/history?index_name=hopper&region_group=US"