Is Pearl (PRL) mining profitable? Formula, dated numbers and calculators

Pearl mining income comes down to four numbers: your hashrate, the network difficulty, the block reward and the PRL price. On October 6, 2026, at $1.14 per PRL, a tuned RTX 5090 (417.9 TH/s, hashrate.no) grossed about $9.56 a day and an RTX 3090 (110 TH/s) about $2.52, before power and fees. Costs come next: electricity or GPU rental, the pool fee and the miner's dev fee. Use the formula below as your own Pearl mining calculator: it shows each input, the values of October 6, 2026, six worked NVIDIA examples, and why the result keeps shrinking.

The formula

Pearl pays a block reward to whoever finds a valid block, so over time your share of the blocks equals your share of the total work. Difficulty measures the work one block needs: in Pearl's code, difficulty D means about D × 2⁴⁸ hashes (2⁴⁸ is 281 trillion). Speed is given in TH/s, trillions of hashes per second; 1 PH/s is 1,000 TH/s and 1 EH/s is a million TH/s.

PRL per day = hashrate (H/s) × 86,400 × block reward / (difficulty × 2^48)

In words: your daily hashes divided by the hashes needed per block give the blocks you find per day on average. Multiply by the reward for PRL, then by the price for dollars. Block time is not in the formula. When blocks come faster than the 194-second target (3 min 14 s), you still earn the same per hash, and difficulty climbs until blocks slow down.

The simpler view: PRL per TH/s per day

With 1 TH/s (10¹² H/s) and the values of October 6, 2026:

1e12 × 86,400 × 2,277.05 / (34,812,805 × 281,474,976,710,656) = 0.0201 PRL

Each TH/s earned 0.0201 PRL per day, or $0.0229 at $1.14, before fees. A 300 TH/s card earned about 6.02 PRL ($6.87). To update the yield later: 0.0201 × (34.81 million ÷ new difficulty) × (new reward ÷ 2,277.05). Or run it with your own numbers:

python3 -c "H=300e12; R=2277.05; D=34812805; print(H*86400*R/(D*2**48))"

Where to get each input

• Difficulty: the official explorer or WhatToMine's coin 469 data.
• Block reward: 2.1e9 × 650,226 / ((h + 650,226) × (h + 650,225)) PRL at height h, from validate.go (tokenomics).
• Price: CoinGecko's Pearl page, not the Perle token that shares the ticker (fake PRL tokens).
• Your hashrate: your pool's 24-hour average, or the hashrate by GPU benchmarks.

Inputs on October 6, 2026

Data as of October 6, 2026, ~17:00 UTC. Sources: explorer.pearlresearch.ai, WhatToMine, CoinGecko, each pool's site, each miner's README.

Hashrate trackers disagree mostly because they average over different windows or use their own estimators. WhatToMine showed about 59.4 EH/s. The node's own estimate (getnetworkhashps), relayed by HeroMiners and other pools, showed about 69.9 EH/s at block 124,165 (~17:20 UTC). Difficulty is the hard number, so the formula uses it directly. MiningPoolStats shows each pool's share.

Worked examples

hashrate.no lists tuned, verified user benchmarks; WhatToMine uses pool-reported hashrates. They differ by about 5% for four of these cards, 18% for the RTX 3090 and 40% for the RTX 5090. Figures are gross: before electricity and fees.

Data as of October 6, 2026, ~17:00 UTC: yield 0.0201 PRL per TH/s per day, price $1.14. Sources: hashrate.no, WhatToMine, official explorer, CoinGecko.

Now subtract electricity: watts × 24 ÷ 1,000 × your price per kWh. The break-even power price is where the card earns nothing.

Data as of October 6, 2026, ~17:00 UTC. Net = gross at hashrate.no benchmarks minus electricity, before fees. Power is the card's draw from hashrate.no; the rest of your system adds to the bill.

Fees come off the gross: the pool fee, plus the miner's dev fee (a share of hashing time paid to its developer). At 2% + 2%, an RTX 5090's $9.56 becomes about $9.19; check the dated pool comparison. Tuning helps: newer hashrate.no benchmarks reach 132–137 TH/s for an RTX 3090 and 323 TH/s for an RTX 4090. The best AMD card on hashrate.no, the RX 9070 XT, does 0.29 TH/s per watt, about a third of a tuned RTX 4090 (0.93).

Hardware cost is not in these tables. For payback, divide what you paid for the card by its daily net, and remember that the daily net shrinks.

Why revenue keeps falling

Revenue per TH/s fell about 85% between the late-May peak and October 6, 2026. Three forces push PRL per TH/s down, and new supply weighs on the dollar side. Revenue per TH/s rises again only if the PRL price rises or hashrate leaves the network, as happened in early August 2026.

• Difficulty is still catching up. Blocks averaged about 165 s in the 24 hours before the snapshot, faster than the 194 s target. Pearl retargets at every block with a smoothing filter (WTEMA, a weighted moving average with a one-week time constant), so difficulty rose 2.6% in those 24 hours, to 34.81 million. At constant hashrate it must rise about 18% more before blocks return to 194 s, cutting revenue per TH/s by about 15% at the same price.
• The reward shrinks every block. No halvings: 2,277.05 PRL at block 124,157, falling about 0.006 PRL per block. At 194 s blocks it will be about 1,555 PRL a year later, 32% lower.
• More hashrate joins. New cards and faster miners raise difficulty. A June 2026 preprint (single author, under review) put the network at about 24 EH/s; on October 6, difficulty implied about 59.5 EH/s.
• Supply roughly doubles within a year. About 336.7M PRL had been mined by October 6, 16.0% of the 2.1 billion cap. At 194 s blocks supply reaches about 642.6M in 12 months. The ~1.19M PRL mined per day (~$1.36M at $1.14) equaled about 46% of the $2.98M reported 24-hour exchange volume. Whether demand absorbs that supply is unknown.

From $33.80 a day to today

On May 31, 2026, Tom's Hardware reported that hashrate.no had listed an RTX 5090 at about $33.80 a day, then $17.19, without the hashrate or price behind them. Some guides still repeat these figures undated. On October 6, hashrate.no's own estimate was about $8.1–8.3 a day, about 75% lower; our difficulty-based figure was $9.56. The drop per card is smaller than per TH/s because RTX 5090 miner speed roughly doubled. It went from about 220 TH/s in May (Kryptex's sample rig, our estimate) to 417.9 TH/s today (hashrate.no).

Difficulty from the official explorer, reward from the emission formula, daily prices from hashrate.no in May (CoinGecko data starts June 22) and CoinGecko after. Gross, before fees. The last column is hypothetical: no card reached 418 TH/s in May.

Difficulty peaked near the July 23 dense-only fork as the price fell to about $0.23. It then fell to about 21.9 million by August 3 (block 95,000) as hashrate left. It stayed near that level through mid-August (21.8 million at block 100,000), after the rank-penalty soft fork of August 6 cut the speed of miners that did not update. The September price rise lifted revenue again. Neither price nor difficulty can be predicted.

Rent or buy?

Renting by the hour avoids the hardware cost and the risk that a future fork shuts your card out. But the rental price replaces your power bill and is usually higher. Below are the cheapest reliable on-demand offers on Vast.ai, a GPU rental marketplace, against the gross revenue above.

Data as of October 6, 2026, ~17:00 UTC: cheapest on-demand single-GPU offers with reliability of 0.95 or more. Margin = gross revenue at $1.14 minus rental, before fees. Break-even = daily rental ÷ PRL per day. The H100 row uses MiningBoard's 700 TH/s, the only published figure, which looks unreliable. Sources: Vast.ai, hashrate.no, WhatToMine.

The cheapest offers often come from unverified hosts. The cheapest verified hosts cost more (RTX 5090 about $0.41 an hour, RTX 4090 about $0.35, RTX 3090 about $0.14), which turns every margin in the table negative. Fees of 2–4% raise each break-even price by 2–4%.

On October 6, 2026, renting was roughly break-even or a small loss. It only pays when the PRL price rises faster than rental prices and difficulty. Buying removes the rental markup but adds the purchase price, electricity and fork risk. The proposed FP8 upgrade (PIP-3) was still a draft with no activation height on October 6. Its reference miner accepts RTX 50 and data-center cards, but not RTX 30 or RTX 40. On September 19, 2026, the SRBMiner developer warned that an upcoming upgrade could make consumer GPUs "pretty much useless". See hard forks.

Solo vs pool

In the long run, solo and pool mining pay the same expected amount, minus fees. What differs is variance: how irregular your income is.

In a pool, your miner sends shares (proofs of partial work), the pool finds blocks with everyone's hashrate and pays you by your share of the work, minus its fee. PPS+ (pay per share plus, Kryptex) pays a fixed amount per share whatever the pool's luck. PROP (proportional: HeroMiners, LuckyPool) and PPLNS (pay per last N shares: Pearl Fortune, Pearlski and others) split each block among recent shares. Income then follows the pool's luck.

Solo, you keep the whole block reward, 2,277.05 PRL on October 6, 2026, but only when your own hardware finds a block. The expected time between your blocks is difficulty × 2⁴⁸ ÷ your hashrate. For one RTX 5090 at 417.9 TH/s:

34,812,805 × 2^48 / 417.9e12 = 23.4 million seconds ≈ 271 days

Finding blocks is random, so the chance of at least one block is 1 − e^−days/271. That is about 10% in 30 days, 28% in 90 days and 74% in a year. It is lower in practice because difficulty keeps rising.

Data as of October 6, 2026, at a difficulty of 34.81 million. Averages only: real waits vary widely.

Pool-run solo modes include Kryptex SOLO (1% as of October 6, 2026, with a solo: prefix before your address) and 2Miners SOLO (1.5%, which reported a 7.3% orphan rate: found blocks that did not stay in the chain). You can also run your own pearld node with --miningaddr and a miner that supports it (BzMiner documents this). The official vLLM miner mines solo to your own node, but only on H100 and H200 cards. New coins can be spent after 100 blocks (about 4.6 hours at recent block times, 5.4 hours at the 194-second target). The mining guide covers both setups, and the wallet guide shows how to get a prl1 address to mine to.

Calculators compared

Each calculator makes its own assumptions, so the same card can show different revenue. On October 6, 2026:

Data as of October 6, 2026, ~17:00 UTC. Sources: each calculator's Pearl pages.

The method on this page needs no hashrate estimate, only difficulty and block reward, both checkable on the official explorer. Your pool dashboard is the final test: after a few days, compare the PRL you received with the formula's prediction for your average hashrate.

Frequently asked questions