Goldshell SC5 Pro II 14 TH/s Blake2B Siacoin Miner

14 TH/s on the Blake2B-Sia algorithm from 3,300W, which is about 236 J/TH, with Goldshell rating hashrate at roughly 3 percent tolerance and power at 5 percent. It is air cooled by two fans at a published 55 decibels, connects over Ethernet, and operates between 5 and 35 degrees Celsius at 10 to 65 percent humidity. Goldshell shipped it from January 2026, making it the newest machine in the Blake2B field. Across that field it places tenth on efficiency, which is honest rather than flattering, and the reasons to choose it are the noise figure and the generation rather than the efficiency ranking.

Not as far as any manufacturer source shows, and this needs saying because several retail listings claim otherwise. The figures those listings quote, 8.8 TH/s at 2,020W, belong to the original SC5 Pro rather than to the Pro II, and at least one distributor reproduces the older machine's entire description on its Pro II page. Unless Goldshell documents a second mode for this model, treat it as a single mode machine running at 14 TH/s and 3,300W. If an adjustable draw matters to you, the original SC5 Pro genuinely does offer one, and in its low power mode it reaches about 229 J/TH, which is better than the Pro II's 236 at default.

This is the comparison that decides the purchase, and it is unusually clean because the two machines draw almost the same power. The SC6-SE gives 17 TH/s from 3,330W at about 196 J/TH and 80 decibels. The SC5 Pro II gives 14 TH/s from 3,300W at about 236 J/TH and a published 55 decibels. So for a 30 watt difference you choose between 3 more terahash and 25 fewer decibels. On economics the SC6-SE wins outright, with more output and better efficiency, and over several years that efficiency gap compounds. The SC5 Pro II wins when your site cannot absorb 80 decibels, in which case the SC6-SE is not a cheaper option you are declining but a machine you cannot deploy.

It is a clear improvement on both counts. The SC5 Pro produces 11 TH/s from 2,820W at about 256 J/TH; the Pro II produces 14 TH/s from 3,300W at about 236. So you get 27 percent more hashrate for 17 percent more power and roughly 8 percent better efficiency. The trade off is that the older machine has a documented low power mode at 8.8 TH/s and 2,020W which the Pro II does not appear to offer, and in that mode the SC5 Pro reaches about 229 J/TH, beating the Pro II's default figure. So if you want maximum output per unit take the Pro II, and if you want the lowest energy per terahash or an adjustable draw, the original SC5 Pro in low power mode is the better tool.

The iBeLink BM-S3 and BM-S3+ are stronger machines, at 19 TH/s from 3,100W and about 163 J/TH and 25 TH/s from 3,400W and about 136 respectively, so both beat the SC5 Pro II on output and efficiency at similar or lower power. They also run at 75 decibels, which is quieter than the SC6-SE but still 20 decibels above this machine's published figure. Against the smaller iBeLink models the SC5 Pro II does well, comfortably beating the BM-S1 Max at full power, which manages 12 TH/s at 262.5 J/TH. If you want the most hashrate and the best efficiency in this power class and noise is not your constraint, the BM-S3 is the machine to compare against rather than anything in Goldshell's range.

It does not, once you know what the unit measures. Joules per terahash describes the cost of one algorithm's work, and Blake2B work is not SHA-256 work, so a Bitcoin machine's 13 and a Sia machine's 236 are answers to different questions rather than points on one scale. Nobody mining Blake2B is competing with Bitcoin hardware for blocks. Inside the algorithm the useful spread runs from roughly 136 at the top with iBeLink's BM-S3+ to roughly 346 at the bottom with the BM-S1, and this machine's 236 places it tenth of fourteen configurations currently available.

The published figure is 55 decibels, which appears in Goldshell derived specifications and several distributor listings, and if it holds it makes this the quietest machine in its power class by a wide margin. We would still ask you to confirm it with us before planning a location, because published noise figures for Goldshell's larger machines are inconsistent across retailers and the original SC5 Pro is quoted variously at 55 and 80 decibels depending where you look. On the assumption of 55 decibels, a garage, outbuilding, utility room or workshop is realistic where an 80 decibel machine would not be. What is not in doubt is the heat: 3,300W sheds roughly 11,260 BTU per hour, so ventilation needs planning regardless, and the operating window is 5 to 35 degrees Celsius.

Confirm this with us before ordering. Our specification table shows 220V, and at 3,300W the machine draws roughly 14 amps at 240V and needs its own properly rated circuit. One distributor listing publishes an input range of 100 to 240V for this model, which cannot be correct on a machine of this size, since at 100V it would draw over 30 amps; Goldshell's comparable industrial models run high voltage ranges around 176 to 285V, and the safe assumption is that this one is similar. Do not plan a circuit or order an adapter on the strength of a 100V figure. Setup otherwise follows the standard Goldshell workflow: connect power and Ethernet, locate the miner on your network, open the web interface, change the default credentials, then enter a Siacoin wallet address and a pool that supports Blake2B.

Realistically one: Siacoin. ScPrime runs the same algorithm and will accept the hashrate, but it is a far smaller network and not a meaningful hedge. What you do not get is optionality. A SHA-256 owner can follow the economics between Bitcoin, Bitcoin Cash and a long tail of smaller chains; a Blake2B owner has nowhere to go. Spending this much on a machine tied to a single project is a position on that project, and it should be taken deliberately. Bitcoin, Litecoin and Kaspa are all outside what this hardware can do.

Three inputs decide it and two of them belong to the market: what Siacoin trades at, how much hashrate is competing, and what you pay per kilowatt hour. On a network this size the first two can shift within weeks, which is why any daily figure you find quoted for this model is already historical. At 3,300W with no low power mode to fall back on, your tariff is the dominant variable and you cannot dial the machine back if prices turn, which makes hosting at industrial rates worth pricing alongside the hardware. On the practical details, most distributors state a PSU is included and publish a manufacturer warranty, though both vary by channel and batch, so confirm at order. Published dimensions and weight also differ between sources, with some apparently quoting the shipping carton rather than the machine, so ask us for the figures on the units we ship if the footprint matters to your rack or shelf planning. MillionMiner ships worldwide DDP with customs, duties and taxes handled.