Free volume bot: costed properly
Free software is a real option and worth taking seriously. What it changes is one line of a campaign budget, and not the largest one. The network fees, the failed transactions, the RPC capacity and your own time are identical whether the code cost nothing or a percentage. Here is that arithmetic, worked from measured block data.
What "free" covers, precisely
There are two genuinely free categories. Public repositories, which give you auditable code and no support, examined in open-source volume bots. And free tiers or trials from hosted vendors, which are usually capped low enough to demonstrate the interface rather than run a campaign.
Neither is a trick. Both are simply narrower than the word suggests, and the useful response is to cost the whole campaign rather than the licence.
What costs the same either way
| Line | Value |
|---|---|
| Software licence | 0 SOL |
| Swaps to land | 2,000 |
| Measured failure rate (Pump.fun AMM (post-graduation)) | 23.3% |
| Attempts required | 2,608 |
| Network fees, total | 0.0215 SOL |
| Of which spent on failures | 0.005 SOL |
| Paid RPC capacity | recurring, in fiat |
| Your engineering time | the largest line |
The row worth staring at is the failure spend. That SOL bought nothing, and it is charged identically whether you paid for software or not. Measured failure rates per DEX program, with the sampling method, are on the measurement page.
The RPC line is the one most often omitted from free-tool comparisons. Public endpoints rate-limit under campaign load, so a campaign of any size needs paid capacity. That is a recurring bill in fiat, and it does not appear in any repository README.
Where the break-even actually sits
Stated against our own commercial interest: if you are technically comfortable and intend to run campaigns repeatedly, build it yourself and keep the 2%. The service fee pays for the operational layer, and if you are going to own that layer anyway, you should not also pay someone else for it.
Where hosted genuinely wins is the first campaign, because the learning costs are front-loaded and they are paid in wasted fees rather than in time. Misconfigured compute limits, inadequate retry logic and rate-limited RPC all produce spend with no volume, and those mistakes are most likely on the first attempt.
What neither option changes is the underlying economics. Full model in the cost breakdown.
What to check before trusting anything free
- Key handling. The only risk that cannot be undone. Search every network call in the code and confirm no key material leaves the machine.
- Hardcoded recipients. An embedded address receiving a share of trades is a fee you did not agree to, and it appears in the wild.
- The cap on a free tier. Usually low enough to show the interface working, not to run a campaign. Find the number before planning around it.
- Maintenance recency. DEX program interfaces change. Code that worked six months ago may fail in ways that look like bad luck rather than breakage.
If you cannot audit code that handles keys, the honest conclusion is that free is not available to you at an acceptable risk, regardless of price.
The paid alternative is priced on the pricing page, and the network cost that applies whichever route you take is measured on the Solana volume bot overview.
There is also a coverage cost that never appears in a repository README. A free script that only speaks Raydium is useless the day your token is still on a Pump.fun curve, and it stops again at graduation when the pool becomes PumpSwap. Tokens whose depth sits in Meteora DLMM or Orca Whirlpools need yet another integration, and routing everything through Jupiter to avoid the work buys the highest failure rate we measure.