Balancer API: what it covers, and a wind-down vote ahead

Most of what gets linked around Balancer is the SDK or the subgraph, yet the GraphQL server behind the Balancer app is public, keyless and detailed, right down to a per-reward-token breakdown of APR.

What follows is what I found by reading that schema live, what a chain-wide index like DexPaprika adds on top of it, how small Balancer's footprint is on every chain I could check it against, and the governance vote that could end the protocol before the year is out. Method at the end, including two numbers that disagree and I am not going to pretend they don't.

What the Balancer API actually gives you

Introspecting the schema at api-v3.balancer.fi/graphql returns 25 top-level query fields. Grouped by what they do:

It answers without a key, over plain POST:

The pool object alone carries more than most single-DEX APIs bother with: isPaused, isInRecoveryMode, holdersCount, aggregateSwapFee, and an aprItems array typed across 19 reward categories, from plain swap fees to Aura and Merkl incentive campaigns. If the job is "describe a Balancer pool" or "what is this LP earning," this is the right tool, and a chain index will not do it better.

What a chain-wide index adds instead

Two things are missing, and both are structural. There is no OHLCV endpoint, only poolGetSnapshots, a daily record of liquidity, share price and volume you would have to turn into candles yourself. And there is nothing outside Balancer. The API's GqlChain enum lists 17 values, 16 live networks plus a Sepolia testnet, but every one is Balancer's own deployment. Ask it about Uniswap or Curve on the same chain and no query answers.

That is the case for a multi-DEX index: Balancer's data only ever describes Balancer. A schema that returns the same shape for every DEX on a chain is what lets you compare one against nine others without nine integrations, which is where DexPaprika's value sits for Balancer specifically.

Balancer's share of DEX volume, 23 September 2026

DexPaprika indexes Balancer under two ids: balancer_v2 on Arbitrum, Ethereum and Polygon, and balancer_v3 on Plasma, its only v3 deployment in this set. Measured against every DEX row DexPaprika carries for that chain, from a single snapshot taken at 22:17 UTC:

Add the four rows up and Balancer's entire DexPaprika-indexed footprint comes to about $4.0 million a day, against $69.5 million of total Polygon DEX volume alone that same day. Polygon is the outlier by a wide margin: everywhere else Balancer is under a fifth of one percent of the chain.

A protocol voting on whether to keep running

Two events matter for anyone integrating Balancer now. On 3 November 2025, at around 07:40 UTC, an attacker exploited a rounding flaw in the scaling math of Balancer v2's Composable Stable Pools, with losses of over $120 million. Certora's and OpenZeppelin's technical write-ups place the bug in v2's stable-pool math, and Certora's states that v3 was not exposed to it.

Then, on 14 September 2026, a proposal was posted on Balancer's own forum for an "orderly winddown": no new business development, pools moved to withdrawals only on 30 October 2026 where the contracts allow it, and the DAO closed to the extent legally and practically possible. Its timeline puts the Snapshot vote at 25 to 29 September 2026, two days after the measurements in this article. If it passes, from 1 November the infrastructure steps down to a simplified withdrawal interface, the subgraph coverage it needs and public documentation, kept available through both rounds of a treasury distribution to BAL holders, the first opening at the end of May 2027. Not everyone in the thread agrees: a fork was pitched there as another direction the day after the proposal went up.

Where the two volume readings disagree

I queried Balancer's own protocolMetricsChain nine minutes after the DexPaprika snapshot above, and the two do not agree on any of the four chains:

Both readings are 24 hour swap volume, both live and keyless, dated within ten minutes of each other. On Ethereum, Arbitrum and Plasma, Balancer's own figure runs higher than ours. On Polygon it runs the other way, by a factor of eight. I have not resolved which counting rule produces the gap and will not guess at it here. If a Polygon-specific Balancer number matters to what you are building, pull both live.

Candles and swaps for a Balancer pool

The busiest Balancer V2 pool on Polygon that day was a USDC/USDT0 pair doing $595,957 of the volume above, on liquidity of under $9,000. Both calls below work without a key.

Hourly candles, which nothing in Balancer's schema produces directly:

Each row carries time_open, time_close, open, high, low, close and volume. The individual swap feed behind it:

Rows carry token_0_symbol, token_1_symbol, both amounts, price_0_usd and price_1_usd, created_at and factory_id, the last one pointing at the Balancer Vault contract that every V2 pool routes through. The same two calls, same shape, work against any of the other DEXes on Polygon, which is the whole point of indexing a chain instead of a single protocol.

No key is needed to try this. A free key from console.dexpaprika.com raises the rate and the quota, current numbers for every tier live on the pricing page since they move, and it is worth knowing before you build that free-tier data can run up to 60 seconds behind, with real-time reserved for the paid tier.

Common questions

Does Balancer have an API?

Yes. A public GraphQL server at api-v3.balancer.fi, documented at docs.balancer.fi under Data and Analytics, with 25 top-level queries covering pools, prices, swap and LP events, protocol metrics and veBAL governance.

Do I need an API key for Balancer?

No. Every query I ran on 23 September 2026 answered with no key and no authentication header of any kind.

Is there a Balancer v3 API?

The same GraphQL server covers both versions; pools carry their protocol version in their type and pool id rather than living on a separate host. DexPaprika distinguishes them as separate dex ids, balancer_v2 and balancer_v3.

Does the Balancer API give OHLCV or price candles?

No. There is no open, high, low, close endpoint. poolGetSnapshots returns daily liquidity, share price and volume, which is a time series but not a candle, and tokenGetHistoricalPrices covers token prices rather than pool prices.

Is Balancer shutting down?

Not decided yet. A proposal to wind the protocol down and return its treasury to BAL holders is up for a Snapshot vote from 25 to 29 September 2026. The proposal's withdrawals-only date is 30 October 2026; the rest of what happens if it passes is in the section above.

Should I use Balancer's API or a DEX aggregator API for Balancer data?

For anything scoped to Balancer alone, pool detail, APR breakdown, governance, LP events, its own API is the more complete source and the one Balancer itself keeps current. Reach for a chain-wide index like DexPaprika when the question spans more than one DEX, needs OHLCV, or has to keep working unmodified if Balancer's own infrastructure changes.

How these numbers were measured

The Balancer API schema and every field name came from a live introspection query against api-v3.balancer.fi/graphql on 23 September 2026. The query-field count of 25 and the pool, event and APR-type field lists are read directly off that live schema.

DEX volumes and shares come from a snapshot of GET /networks/{network}/dexes for Arbitrum, Ethereum, Polygon and Plasma, taken at 22:17 UTC. Each chain's share is Balancer's row divided by the sum of every DEX row DexPaprika returned for that chain in the same snapshot: a share of that chain's indexed DEX volume specifically. Balancer's own swapVolume24h figures came from protocolMetricsChain at 22:27 UTC the same day and are quoted as returned, no adjustment.

The busiest-pool figures and the OHLCV and transaction calls are quoted from live DexPaprika responses for pool 0x74d5e70a8db6281ef5341d609725b1c7bbf5ebc6 on Polygon. The exploit date, time and mechanism come from Certora's and OpenZeppelin's independent technical write-ups, cross-checked against each other. The winddown facts, including the vote dates and the 30 October withdrawals-only date, are read from the proposal's own timeline on forum.balancer.fi. Balancer forks that DexPaprika lists under their own names (Beets on Sonic, Beethoven X on Optimism, BeraSwap on Berachain) are not counted as Balancer here.

Every volume and share figure is a 24 hour window closing at measurement time, so expect movement, and the vote outcome was not yet known when this was written.

Full REST reference at docs.dexpaprika.com. Balancer's own API is documented at docs.balancer.fi. For the wind-down proposal itself, the source is Balancer's governance forum, and the wider pattern of DeFi attacks is covered in DeFi exploits in 2026.

The same API-coverage question for two other DEXes: what the Uniswap API returns, and what Raydium's API covers on Solana.