Blocknative의 가스 API가 2일 후에 어두워집니다 — 5분 마이그레이션 체크리스트

Blocknative’s gas estimation API shuts down June 19, 2026. If your stack still calls
api.blocknative.com/gasprices/blockprices, those calls start failing this week.

This is a practical checklist, not a pitch — two of the three options below don’t involve us.

1. Find every callsite

grep -rn "api.blocknative.com" . --include="*.{js,ts,py,go}"

Enter fullscreen mode Exit fullscreen mode

Don’t forget env vars and config files: grep -rn "blocknative" .env* config/.

2. Pick a replacement

Option A — your chain’s official endpoint (free). Polygon has the
Gas Station. Decentraland’s governance backend
migrated this way last week — clean PR
worth copying if you’re Polygon-only.

Option B — your RPC provider. eth_feeHistory is enough for most apps; Alchemy/Infura
tiers include it. You lose the confidence-band format and write the percentile math yourself.

Option C — a drop-in with the same schema (ours). If you don’t want to touch parsing code:
same blockPrices JSON shape, no API key, no signup. Honest disclosure: we’re small, the free
tier is 100 calls/day per IP, and beyond that it’s $0.001/call via x402. Live right now on
Ethereum, Optimism, Polygon, Base, Arbitrum (5/5 verified as this post was generated):

curl "https://logicnodes.io/gasprices/blockprices?chainid=1"

Enter fullscreen mode Exit fullscreen mode

One-line swap, Python/ethers.js snippets, and chain coverage caveats:
logicnodes.io/blocknative

3. Verify before June 19

Whatever you pick, deploy behind a feature flag and compare both sources for a day:
log maxFeePerGas deltas; anything within ~10% at confidence 99 is fine for inclusion.

Caveats

• We cover 5 chains, not Blocknative’s 40+. If you need more, Option B is your friend.
• Confidence values are recomputed from eth_feeHistory over 100 blocks — methodology differs from Blocknative’s mempool model; absolute values track within noise on quiet chains but diverge during gas spikes.
• This post was generated by our infrastructure and the endpoint health-check above ran at publish time. Numbers in our docs are live, not cached marketing claims.