Bitquery vs Anaxer for pump.fun data: four workloads, priced

Quick answer

Bitquery is a genuinely strong pump.fun data platform - bonding curve progress, a dedicated migration stream, gRPC, Kafka, and 40+ chains. Its weakness is not coverage, it is the meter. Streaming is billed three ways at once, and the gigabyte meter is the one a pump.fun firehose trips: we measured that firehose at ~235 GB a month across pump.fun and PumpSwap, against 5 GB on Bitquery's $99 plan and 50 GB on its $299 plan.

So the answer depends on what you are building. Below, four real workloads, priced on both.

Why this comparison is shaped differently

Our Helius and Birdeye comparisons are straight head-to-heads, because in both cases one product is broadly cheaper for pump.fun and the tables say so.

Bitquery does not resolve that cleanly. It covers pump.fun properly - arguably in more depth than we do - and its pricing is efficient for some workloads and punishing for others. A single "which is cheaper" verdict would be false. So this page prices four concrete jobs instead, and Bitquery takes one of them.

What are Anaxer and Bitquery, exactly?

Bitquery is a blockchain data platform covering 40+ chains through a single GraphQL schema, with real-time WebSocket subscriptions, gRPC (CoreCast), Kafka topics at the Enterprise tier, and mempool streams. Its pump.fun coverage is deep and deliberate: live trades, new token launches, bonding curve progress with a published completion formula, and a dedicated migration subscription for tokens graduating to PumpSwap.

Anaxer is a Solana-only real-time event API with five channels - creations, graduations, trades, prices, and transfers (coming soon) - delivered as pre-parsed JSON over WebSocket and REST, on flat monthly plans.

The important structural difference is what you pay for. Bitquery prices consumption: every API call spends points, every connected minute spends stream-minutes, every byte pushed spends gigabytes. Anaxer prices access: your plan sets which channels you can open and how many concurrent connections you get, and volume is never counted.

That difference is neutral in the abstract. It only becomes an advantage or a liability once you know your workload - which is the whole point of what follows.

How much data does pump.fun actually move?

Every cost estimate below rests on this, so here is the measurement rather than a guess.

We took trade counts from a full 24-hour production window (22-23 September 2026) and payload sizes from live samples on the wire:

The other channels are trivial by comparison - prices ~17 GB/month, creations ~0.4 GB, graduations ~0.02 GB. Trades are the entire cost story.

The single most useful number here is that PumpSwap carries 3.4x the trade volume of the bonding curve. Most pump.fun activity happens after graduation, on the AMM. Anyone sizing a pump.fun workload from launches or bonding-curve trades alone will underestimate their bandwidth by more than four times - and on a per-gigabyte meter, that error is the whole bill.

Two things follow. First, trades dominate - they are two orders of magnitude more voluminous than launches. Second, payload shape is a cost lever when someone is charging you per byte. Our trade event is 584 bytes. Bitquery's documented pump.fun subscription returns a full Currency object - name, symbol, mint address, decimals, fungible flag, URI - on both legs of every trade, which lands closer to 1.2-1.5 KB. Same information content, two to three times the meter reading.

You can trim that by requesting fewer fields, and you should. But it means the honest answer to "what does this cost on Bitquery" is a range, not a number.

Four workloads, priced

Pricing verified 23 September 2026 against the public Bitquery pricing page and the Anaxer pricing page. Bitquery prices are shown month-to-month; annual billing is 20% less but prepays twelve months.

Workload 1: watch every new pump.fun launch

You want a live feed of every token created - roughly 21,000-33,000 a day, but tiny in bytes (~0.4 GB/month).

Bitquery has no cheap entry for this. The $49 Personal tier excludes WebSocket entirely, and its licence forbids commercial use - so a hobbyist bot that might one day earn money cannot sit there either. Anaxer wins by $99.

Workload 2: trades for a 20-token watchlist

You track a fixed basket of tokens and want every trade on them. Filtered server-side, so volume is modest and comfortably inside 5 GB.

This is a genuine draw, and on balance Bitquery gives you more for the money - the same $99 buys 40+ chains, mempool streams, bonding curve progress as a ready field, and a full GraphQL schema rather than five fixed channels. Our 10-token cap on the $39 tier is the constraint that pushes you up, and that is a limitation of our plan, not of theirs.

If your watchlist is 10 tokens or fewer, Anaxer Starter at $39 does it and Bitquery still has no sub-$99 streaming option. Between 10 and 100, they are level on price and ahead on breadth.

Workload 3: the full chain-wide firehose

Every pump.fun and PumpSwap trade, unfiltered - the workload for a sniper, a market-wide dashboard, or any product that cannot know in advance which token will matter. This is ~235 GB/month.

Here is where the model breaks down, and the arithmetic is worth walking through.

Pro ($99) is not an option. Its 5 GB allowance is roughly 2% of what this workload needs - about 15 hours of trades. It is not a tier you can top up into; it is the wrong tier.

Scale ($299) does not cover it either. Its 50 GB is about a fifth of the month. You need roughly 185 GB more, which at Bitquery's volume-pack rates is two 100 GB packs - around $600/month on top of the $299 base. Call it ~$900/mo month-to-month, or ~$719 on annual billing if you prepay twelve months.

And that is the optimistic case. It assumes you trim the GraphQL selection set down to roughly our payload size. Run the subscription as Bitquery documents it - full Currency objects on both legs of every trade, at 1.2-1.5 KB - and the same firehose is over 500 GB a month, which is a 500 GB volume pack and a bill near $1,300/mo, or an Enterprise conversation.

If you only need the bonding curve and not PumpSwap - pre-graduation sniping, say - the picture is far kinder: ~54 GB/month, which is Scale plus a single 5 GB top-up, about $349/mo. Still three and a half times our $99, but a different order of problem.

Stream-minutes, incidentally, are a non-issue here - four streams running 24/7 is ~173,000 minutes against Scale's 2,000,000. Bitquery's own cost calculator tells you "stream-min binds first." For pump.fun that is simply not true; gigabytes bind first, and by a wide margin. The calculator is tuned for lighter workloads than this one.

Workload 4: backfill and historical analysis

You need to reconstruct what happened - chart a token's history, backtest a strategy, or fill a gap after downtime.

How much this matters depends entirely on how far back you need to look.

Anaxer does keep history - the same events are queryable over REST for 7 days on trades and 90 days on creations and graduations. For gap recovery after a disconnect, or analysis over the last week or last quarter of launches, that is the whole job and it is included in the plan you already have.

Past that, Bitquery wins and the margin grows with the lookback. Their self-service trades window is 30 days against our 7, so anything between one and four weeks old is theirs. Beyond 30 days, only their paid archive packs reach it at all. If you are backtesting a strategy over six months, we are not a candidate and no amount of streaming economy compensates.

Worth knowing about their model, though: no self-service plan includes historical data. Personal, Pro and Scale are all real-time windows only - 30 days for Solana trades, 8 hours for transfers. Archive is a separate per-chain purchase. So "Bitquery has history" is true, but it is a line item, not something the $99 plan quietly includes.

Where does Bitquery genuinely beat us?

Beyond backfill, several things, and pretending otherwise would waste your time.

• Breadth. 40+ chains against our Solana-only. If you may ever need Ethereum, Base, BSC or Tron, that is decisive.
• Bonding curve progress as a ready-made field. Bitquery publishes curve completion with a documented formula, so you just read it. The same signal is derivable from our data - marketCapUsd on the prices stream tracks a pre-graduation token toward its threshold, and the trades stream gives you cumulative buy volume on the curve - but you configure and maintain that yourself. If you want to catch tokens at 90% rather than 100%, Bitquery hands it to you and we make you build it.
• Transport options. WebSocket, gRPC (CoreCast), Kafka at Enterprise, and Shred-level streams that capture transactions as they reach validators - potentially ahead of block finalisation, and so ahead of our sub-450 ms parsed delivery. If you are competing on latency at the edge, that matters.
• Mempool streams. We have no equivalent.
• Query flexibility. GraphQL lets you shape the response precisely. Our channels return a fixed shape. That flexibility is also the mitigation for their gigabyte meter.
• A trial that actually streams. Bitquery's 7-day trial includes full live WebSocket access, which most providers refuse. It is small - 17 stream-minutes and 0.2 GB - but it is real, and rarer than it should be.

So which should you pick?

Pick Bitquery if you need history or backfill, more chains than Solana, bonding curve progress without deriving it yourself, mempool data, or the lowest achievable latency via gRPC and Shred streams - or if your workload is a filtered watchlist, where the $99 tier comfortably covers you and buys more breadth than ours does.

Pick Anaxer if you want the chain-wide pump.fun firehose at a price that does not move with volume, you want launches and graduations streaming on a free plan with no trial clock, or you simply do not want to model three meters to forecast a bill.

The honest summary is that these products are priced for different shapes of demand. Bitquery is efficient when you know exactly what you want and ask for precisely that, across many chains. It gets expensive when you want everything from one very loud source. We are the reverse: narrow, fixed, and indifferent to volume. Neither is the better engineering decision in the abstract - it depends on which of the four workloads above is yours.

Next steps

The cheapest way to settle this is to measure your own workload rather than trust either of our marketing pages. Our creations and graduations streams are free and unmetered, so you can log payload sizes and event rates against your own filters and work out what those bytes would cost you elsewhere. Bitquery's 7-day trial lets you do the same from their side.

If you are weighing raw firehose decoding against parsed events more generally, read raw gRPC or parsed events. For production connection handling, see the Solana streaming guide. When you are ready, explore the pump.fun API or get a key.