Bitcoin puzzle transaction ~32 BTC prize to who solves it

[analyticnomad]

...

What do you mean by "a private miner"? A miner with a private aka non-public mempool or what else?

I'm not aware that ViaBTC has a public service with a private mempool.

You also don't say anything about which puzzle solution's transaction you are talking about. A withdrawal of coins with a solution for puzzle #135 can be broadcasted to public mempools. A withdrawal of coins with a solution for puzzle #71 needs to be broadcasted non-publicly otherwise bots can steal the public key from the transaction in public mempools and find private key in seconds and then the RBF transaction replacement war begins.

This topic is always repeated. Look for another topic such as how to provide some GPUs almost for free. Regarding the 71-bit puzzle, I was able to reduce it to 64-bit. The biggest challenge to solve puzzle 71 is to provide at least 5090x10 to find the key within months.

How did you reduce puzzle 71 to 64?? Having only the address, since Public Key → SHA-256 Hash means you lose the original structure of the public key, then SHA-256 Hash → RIPEMD-160 loses the length and bit uniqueness. The result is shortened from 256 bits to 160 bits (20 bytes). So based on the address, it is impossible to reduce it — the only reduction is the disclosed range in which the address is located.

The vulnerability I discovered is difficult to find, so the creator thought no one would be able to access it, but I did. He mentioned in his post that there's no definitive pattern, but there are two possibilities: either he wasn't aware of the serious vulnerability, or he was aware of it but didn't expect anyone to discover it. Regarding image, 71 puzzle, it falls within the red line. In the next step, this is a 100% certain prediction; it can't be outside the red line. Therefore, the area to be scanned is quite large, but fortunately, the maximum area is 65 bits, so I don't have to scan all 71 bits. After some in-depth research, I found another method that allowed me to reduce the 65 bits to approximately 62 bits. This image is just one of dozens I have, and it doesn't represent the complete method yet because the details are much deeper, and I don't want to share them. https://imgur.com/a/fA9O7Aw  

Someone please install an update on this guy’s brain

I think you're the one who needs a brain update , considering your advanced age. My brain is currently in a growth phase, accepting pure and logical analysis. For example, on March 21, 2026, at 02:37:00 AM, you said, "I would have solved the 71 problem in a few months if I had an RTX 5090 graphics card." Even a baby could understand that you wouldn't be able to find the key with a single 5090, even if you were the luckiest person, not even an astronomical stroke of luck would work. You must accept that someone from the new generation found a vulnerability and you didn't.  

A single 5090 has a significantly higher probability of finding the key at random within the next ten minutes than you having "found a vulnerability".

The probability within 10 minutes is not zero, but it is so small that it is practically impossible.  0.0000002% That means 1 chance in 500 million This means you need, on average, hundreds of millions of the same attempt (10 minutes).   |  The comparison you made to me is scientifically illogical; it's merely an offensive response and not a genuine analysis.

The vulnerability is not a matter of "probability" but a matter of "proof".  

No, It's 100% accurate. The statistics favor cracking #71 randomly in ten minutes over you having an exploit.

You for sure DON'T have an exploit/vulnerability but #71 CAN be found randomly.

Because math always wins vs. delusion.

[napros]

Hi there NUCLEAR7.1, does my prediction align with what you have been able to predict

🚀 ADAPTIVE AI RANGE DEPLOYMENT: P71
──────────────────────────────────────────────────
📦 PRECISION: 615926027D0EA7B000 → 64BBFC4FC6D6405000
             Width:  5.29% (62,462,197,061,138,292,736 keys)

📦  BALANCED: 60ECCB38D3D5B4C000 → 65285719700F334000
             Width:  6.61% (78,077,746,326,422,847,488 keys)

📦    SAFETY: 5FDDE840ACC7550000 → 66373A11971D930000
             Width:  9.92% (117,116,619,489,634,287,616 keys)

📦  COVERAGE: 5ECF054885B8F58000 → 67461D09BE2BF28000
             Width: 13.23% (156,155,492,652,845,694,976 keys)

I would be very interested to know if we are both on the same path.

[Cricktor]

Thank god we're back to the usual nonsense here, lazy excessive and non-sensical pyramid quoting and whatnot stupidity else. For a second I thought there could be any progress. Well... stupid me! Where are the prefix worshippers, out on vacation? I kinda miss them (not)!

[NUCLEAR7.1]

Hi there NUCLEAR7.1, does my prediction align with what you have been able to predict

🚀 ADAPTIVE AI RANGE DEPLOYMENT: P71
──────────────────────────────────────────────────
📦 PRECISION: 615926027D0EA7B000 → 64BBFC4FC6D6405000
             Width:  5.29% (62,462,197,061,138,292,736 keys)

📦  BALANCED: 60ECCB38D3D5B4C000 → 65285719700F334000
             Width:  6.61% (78,077,746,326,422,847,488 keys)

📦    SAFETY: 5FDDE840ACC7550000 → 66373A11971D930000
             Width:  9.92% (117,116,619,489,634,287,616 keys)

📦  COVERAGE: 5ECF054885B8F58000 → 67461D09BE2BF28000
             Width: 13.23% (156,155,492,652,845,694,976 keys)

I would be very interested to know if we are both on the same path.

No, your expectations are wrong.

[blankx4729]

It is been 6+months running keyhunt bsgs searching for 135 in random mode on 4 pc's but no luck, I wish if some one like RC or anyone searching 135 in sequential mode share his status so that we can reduce some keys.

Hi ik316m,

Quick math on why BSGS can't crack 135 regardless of runtime:

Optimal BSGS needs sqrt(N) = 2^67 baby-step entries. At 16 B/entry that's ~2.4 ZB of RAM. Impossible. With realistic memory (1 TB across 4 PCs ≈ 2^36 entries), giant steps required = N/M = 2^134/2^36 = 2^98. Even at an unrealistic 100 GK/s aggregate that's ~10^11 years. Random mode doesn't help — same total work.

Kangaroo hits Shoup's lower bound ~2*sqrt(N) ≈ 2^67.5 ops without needing sqrt(N) storage (DPs compress the table). Using SecretAdmirere's 2.7 GK/s figure for RCKangaroo on a 3070 Ti (#13088): ~1700 years on one old GPU, scales linearly — months at 1000 GPUs. Puzzles 120/125/130 were all solved this way.

135's pubkey is exposed, so kangaroo applies. Switch to RCKangaroo (github.com/RetiredC/RCKangaroo) or PSCKangaroo (github.com/pscamillo/PSCKangaroo) — but be realistic: even with kangaroo, a single hobbyist's odds of hitting 135 are essentially zero (unless you happen to be the luckiest person alive). Only large coordinated efforts have a real shot. The point isn't that you'll find it; it's that BSGS doesn't even let you try.

Hi! First, thank you for this excellent tool and for sharing the theory behind it.
I've been experimenting with PSCKangaroo on a GTX 1060 6GB (sm_61, WSL2) and discovered some interesting behavior around PNT_GROUP_CNT that I wanted to share and get your input on.
Findings
Default build (PNT=64, OLD_GPU path):

Speed: ~0.19 GKeys/s
ptxas reports 5200 bytes stack frame, 136 bytes spill stores on KernelA
Concurrent mode WILDs never appear (always 0 checks)

After changing PNT_GROUP_CNT:
PNTVisual SpeedSpillWILDs working?64 (default)0.19 GKeys/s5200 bytes❌160.25 GKeys/s1344 bytes✅ (after fixes)81.20 GKeys/s704 bytes❌ broken distances42.70 GKeys/s384 bytes❌ broken distances23.80 GKeys/s240 bytes❌ broken distances
Lower PNT = lower register spill = more active warps per SM = higher displayed GKeys/s. But values below 16 break the WILD distance calculation in KernelB due to hardcoded constants.
Bugs Found and Fixed
Bug 1 — KangCnt mismatch in OLD_GPU path:
In GpuKang.cpp, GroupCnt was hardcoded to 64 for OLD_GPU:
cpp// Before:
Kparams.GroupCnt = IsOldGpu ? 64 : V45_PNT_GROUP_CNT;
// After:
Kparams.GroupCnt = IsOldGpu ? PNT_GROUP_CNT : V45_PNT_GROUP_CNT;
With PNT=16 and GroupCnt=64, KangCnt = 512 × 64 × blocks = 786432 but kang_ind max = 16 × (512 × blocks) = 196608. This caused kang_type = 3 * kang_ind / KangCnt to always return 0 (TAME), meaning WILDs were never checked — explaining why concurrent mode showed 0.0K/s WILD checks forever.
Bug 2 — OLD_GPU hardcoded PNT=64 in defs.h:
cpp// Line 111 in defs.h:
#define PNT_GROUP_CNT  64  // hardcoded, ignores V45_PNT_GROUP_CNT
Changed to 16 for correct behavior.
Questions
1. Why does lower PNT increase displayed GKeys/s so dramatically?
My analysis: lower PNT reduces the local array sizes in KernelA (Lx[4*PNT], Ly[4*PNT], Ls[2*PNT]), which reduces register spill, which increases active warps per SM. Is this correct?
FactorPNT=64PNT=16Stack frame5200 bytes1344 bytesActive warps/SM~8~16Effective speed0.19 GKeys/s0.25 GKeys/s
2. Is the higher displayed GKeys/s with lower PNT real or misleading?
At PNT=8 I see 1.20 GKeys/s but puzzle 70 takes longer to solve than at PNT=16 with 0.25 GKeys/s. This suggests the displayed speed is real (more kangaroo steps per second) but the algorithm efficiency per step drops because KernelB's distance reconstruction is broken for PNT<16 due to hardcoded constants like / 128 and 8 * g8_ind.
3. For puzzle 135 with limited RAM — is lower PNT better?
Currently running p135 on 1060 with PNT=16, dp=15, 4GB RAM:
HUNT: Speed: 0.25 GKeys/s | Time: 3h 36m
TAMEs: 207M (93% frozen) | 100% WILDs hunting
Checks: 540M | T-W: 8605 | W-W: 1966 | FP: 19765 | 41.5K/s
BSGS [4 thr]: 955 pending, 9403 processed, 0 dropped
The BSGS queue (955 pending) suggests FP rate is too high even at dp=20. Would increasing dp further help, or is this a fundamental limitation of the truncated distance approach at this range?
4. KernelB index math generalization
The KernelB kang_ind calculation has several constants hardcoded for PNT=64:
cppu32 g8_ind = (tind % (32 * PNT_GROUP_CNT / 2)) / 128; // 128 = 4*PNT/2 for PNT=64 only
u32 kang_ind += 8 * g8_ind + gr_ind; // 8 assumes PNT=64
Would it be feasible to generalize these for arbitrary PNT values? This would allow sm_61 GPUs to benefit from lower PNT without broken distance tracking.

[NUCLEAR7.1]

...

What do you mean by "a private miner"? A miner with a private aka non-public mempool or what else?

I'm not aware that ViaBTC has a public service with a private mempool.

You also don't say anything about which puzzle solution's transaction you are talking about. A withdrawal of coins with a solution for puzzle #135 can be broadcasted to public mempools. A withdrawal of coins with a solution for puzzle #71 needs to be broadcasted non-publicly otherwise bots can steal the public key from the transaction in public mempools and find private key in seconds and then the RBF transaction replacement war begins.

This topic is always repeated. Look for another topic such as how to provide some GPUs almost for free. Regarding the 71-bit puzzle, I was able to reduce it to 64-bit. The biggest challenge to solve puzzle 71 is to provide at least 5090x10 to find the key within months.

How did you reduce puzzle 71 to 64?? Having only the address, since Public Key → SHA-256 Hash means you lose the original structure of the public key, then SHA-256 Hash → RIPEMD-160 loses the length and bit uniqueness. The result is shortened from 256 bits to 160 bits (20 bytes). So based on the address, it is impossible to reduce it — the only reduction is the disclosed range in which the address is located.

The vulnerability I discovered is difficult to find, so the creator thought no one would be able to access it, but I did. He mentioned in his post that there's no definitive pattern, but there are two possibilities: either he wasn't aware of the serious vulnerability, or he was aware of it but didn't expect anyone to discover it. Regarding image, 71 puzzle, it falls within the red line. In the next step, this is a 100% certain prediction; it can't be outside the red line. Therefore, the area to be scanned is quite large, but fortunately, the maximum area is 65 bits, so I don't have to scan all 71 bits. After some in-depth research, I found another method that allowed me to reduce the 65 bits to approximately 62 bits. This image is just one of dozens I have, and it doesn't represent the complete method yet because the details are much deeper, and I don't want to share them. https://imgur.com/a/fA9O7Aw  

Someone please install an update on this guy’s brain

I think you're the one who needs a brain update , considering your advanced age. My brain is currently in a growth phase, accepting pure and logical analysis. For example, on March 21, 2026, at 02:37:00 AM, you said, "I would have solved the 71 problem in a few months if I had an RTX 5090 graphics card." Even a baby could understand that you wouldn't be able to find the key with a single 5090, even if you were the luckiest person, not even an astronomical stroke of luck would work. You must accept that someone from the new generation found a vulnerability and you didn't.  

A single 5090 has a significantly higher probability of finding the key at random within the next ten minutes than you having "found a vulnerability".

The probability within 10 minutes is not zero, but it is so small that it is practically impossible.  0.0000002% That means 1 chance in 500 million This means you need, on average, hundreds of millions of the same attempt (10 minutes).   |  The comparison you made to me is scientifically illogical; it's merely an offensive response and not a genuine analysis.

The vulnerability is not a matter of "probability" but a matter of "proof".  

No, It's 100% accurate. The statistics favor cracking #71 randomly in ten minutes over you having an exploit.

You for sure DON'T have an exploit/vulnerability but #71 CAN be found randomly.

Because math always wins vs. delusion.

I have an exploit/vulnerability 10000000000% NOT delusion  The statistics SAY YOU WRONG 100% show me your analysis

[Niekko]

I have an exploit/vulnerability 10000000000% NOT delusion  The statistics SAY YOU WRONG 100% show me your analysis

No exploit/vulnerability in the puzzle. Who tell different are a scammer.

[NUCLEAR7.1]

I have an exploit/vulnerability 10000000000% NOT delusion  The statistics SAY YOU WRONG 100% show me your analysis

No exploit/vulnerability in the puzzle. Who tell different are a scammer.

you abslotly wrong I DONT ASK FOR MONEY IM NOT A SCAMMER. I found a exploit/vulnerability/pattren you dont know what I found you all just talking disrespectfully

[0xastraeus]

I have an exploit/vulnerability 10000000000% NOT delusion  The statistics SAY YOU WRONG 100% show me your analysis

No exploit/vulnerability in the puzzle. Who tell different are a scammer.

you abslotly wrong I DONT ASK FOR MONEY IM NOT A SCAMMER. I found a exploit/vulnerability/pattren you dont know what I found you all just talking disrespectfully

Holy moly bro..

We get it you found your so called "vulnerability" (you and every other person in here) you don't need to keep spewing it out everywhere here and in the btcpuzzle.info telegram channel congrats...

Just stop and work on it on your own.

[nomachine]

I found a exploit/vulnerability/pattren you dont know what I found you all just talking disrespectfully

Yo, the curve don't leak structure like that, fam. Secp256k1, the elliptic curve Bitcoin runs on, got picked with a fat safety margin. That ECDLP is still believed to be straight-up hard. Puzzle private keys get chosen uniform random in their known ranges, like for #71 it's roughly 2^70 to 2^71 minus 1. Ain't no known math relation, bias, or hidden pattern in the public key or address that actually cuts the entropy on a solid key.If somebody had a real way to shave 6 to 9 bits off random instances just from deep research on hex images or points, that would be a massive breakthrough in ECDLP. Real entropy reduction needs actual stats, like showing puzzle keys fall into certain sub-ranges way more than they should across a bunch of examples, with a model that predicts and ain't just curve-fitting one case.Scale don't add up for no breakthrough anyway. Even if you cut it to 62 bits you still got like 4 times 10 to the 18 keys left. At optimistic big GPU or ASIC farm speeds, hundreds of billions keys per second, that's still years on average. All the solved Bitcoin puzzles got beat by optimized brute force, kangaroo, or distributed search on the known bit range, not by cracking the math or finding weird stuff in the generator point or key gen. Bigger ones stay unsolved exactly like the math says for that 2 to the n over 2 complexity.Human brains are pattern finding machines, built for survival not cracking 256-bit random strings. Throw enough visuals, constants like pi, e, golden ratio, digit tricks, or sacred geometry at pure randomness and you'll see whatever you want. This ain't new, it's just numerology with better software. Same bias that got folks picking lottery numbers, doing astrology, and chasing conspiracy numbers.  

[napros]

Hi there NUCLEAR7.1, does my prediction align with what you have been able to predict

🚀 ADAPTIVE AI RANGE DEPLOYMENT: P71
──────────────────────────────────────────────────
📦 PRECISION: 615926027D0EA7B000 → 64BBFC4FC6D6405000
             Width:  5.29% (62,462,197,061,138,292,736 keys)

📦  BALANCED: 60ECCB38D3D5B4C000 → 65285719700F334000
             Width:  6.61% (78,077,746,326,422,847,488 keys)

📦    SAFETY: 5FDDE840ACC7550000 → 66373A11971D930000
             Width:  9.92% (117,116,619,489,634,287,616 keys)

📦  COVERAGE: 5ECF054885B8F58000 → 67461D09BE2BF28000
             Width: 13.23% (156,155,492,652,845,694,976 keys)

I would be very interested to know if we are both on the same path.

No, your expectations are wrong.

Higher or Lower? (more information would be helpful for those who care)

[NUCLEAR7.1]

Hi there NUCLEAR7.1, does my prediction align with what you have been able to predict

🚀 ADAPTIVE AI RANGE DEPLOYMENT: P71
──────────────────────────────────────────────────
📦 PRECISION: 615926027D0EA7B000 → 64BBFC4FC6D6405000
             Width:  5.29% (62,462,197,061,138,292,736 keys)

📦  BALANCED: 60ECCB38D3D5B4C000 → 65285719700F334000
             Width:  6.61% (78,077,746,326,422,847,488 keys)

📦    SAFETY: 5FDDE840ACC7550000 → 66373A11971D930000
             Width:  9.92% (117,116,619,489,634,287,616 keys)

📦  COVERAGE: 5ECF054885B8F58000 → 67461D09BE2BF28000
             Width: 13.23% (156,155,492,652,845,694,976 keys)

I would be very interested to know if we are both on the same path.

No, your expectations are wrong.

Higher or Lower? (more information would be helpful for those who care)

Lower

[Akito S. M. Hosana]

Human brains are pattern finding machines, built for survival not cracking 256-bit random strings. Throw enough visuals, constants like pi, e, golden ratio, digit tricks, or sacred geometry at pure randomness and you'll see whatever you want. This ain't new, it's just numerology with better software. Same bias that got folks picking lottery numbers, doing astrology, and chasing conspiracy numbers.  

You see, children’s imagination has no limits, and neither, apparently, does the adult version when staring at random-looking hex strings. While you’re out here bravely defending the sacred temple of 'it’s just randomness, stop looking,' some of us are busy building rainbow bridges made of cotton candy between secp256k1 and sacred geometry. Why? Because unlike sober brute-force kangaroos, our inner five-year-old still believes that if you stare at the numbers long enough, a glittery unicorn might poop a private key.   

[Niekko]

I have an exploit/vulnerability 10000000000% NOT delusion  The statistics SAY YOU WRONG 100% show me your analysis

No exploit/vulnerability in the puzzle. Who tell different are a scammer.

you abslotly wrong I DONT ASK FOR MONEY IM NOT A SCAMMER. I found a exploit/vulnerability/pattren you dont know what I found you all just talking disrespectfully

Not yet, but I'm sure the finish of this shit will be to try to sell the revolutionary method. 

Do you want to be trusted ? Prove it. Solve the puzzle.

[NUCLEAR7.1]

I have an exploit/vulnerability 10000000000% NOT delusion  The statistics SAY YOU WRONG 100% show me your analysis

No exploit/vulnerability in the puzzle. Who tell different are a scammer.

you abslotly wrong I DONT ASK FOR MONEY IM NOT A SCAMMER. I found a exploit/vulnerability/pattren you dont know what I found you all just talking disrespectfully

Not yet, but I'm sure the finish of this shit will be to try to sell the revolutionary method. 

Do you want to be trusted ? Prove it. Solve the puzzle.

You're right, but I won't sell it now. If I manage to solve the puzzle someday, I'll sell it then because it will truly be valuable.

[Niekko]

You're right, but I won't sell it now. If I manage to solve the puzzle someday, I'll sell it then because it will truly be valuable.

Someone already answer you about the dreams:

if you stare at the numbers long enough, a glittery unicorn might poop a private key.   

[NUCLEAR7.1]

You're right, but I won't sell it now. If I manage to solve the puzzle someday, I'll sell it then because it will truly be valuable.

Someone already answer you about the dreams:

if you stare at the numbers long enough, a glittery unicorn might poop a private key.   

This pattern that I see gives a recurring structure, and it can be verified, and I did that, and the results were all around all the keys. It is not an illusion. I have all the evidence. Nothing is impossible here if the puzzles were created by a person who goes to the toilet twice a day As everyone does and He breathes the same oxygen that you breathe.  Yes, the puzzles are completely random. But there's a vulnerability.  I could convince everyone with a single click by posting a 7-page PDF, and everyone would be convinced and resort to my method, and the mystery would be solved within months, but I won't do that.

[Niekko]

Yes, the puzzles are completely random.

it's not random. it's a Hierarchical Deterministic Wallet cutted on MSB and with first byte adjusted for be into the range. Maybe only the first byte it's choosed randomly (or maybe not  )

Yes, the puzzles are completely random.

it's not random. it's a Hierarchical Deterministic Wallet cutted on MSB and with first byte adjusted for be into the range. Maybe only the first byte it's choosed randomly (or maybe not  )

Finally, you understand what I mean. Thank God, you answered yourself, haha  

Nothing change, no exploit/vulnerability anyway.

[0xastraeus]

Yes, the puzzles are completely random.

it's not random. it's a Hierarchical Deterministic Wallet cutted on MSB and with first byte adjusted for be into the range. Maybe only the first byte it's choosed randomly (or maybe not  )

Finally, you understand what I mean. Thank God, you answered yourself, haha 

Nothing change, no exploit/vulnerability anyway.

Guys, just stop feeding into his delusion. He found one, something that no other mathematician or cryptographer could ever do in the last decade.

Just leave it at that and move on.

[NUCLEAR7.1]

Yes, the puzzles are completely random.

it's not random. it's a Hierarchical Deterministic Wallet cutted on MSB and with first byte adjusted for be into the range. Maybe only the first byte it's choosed randomly (or maybe not  )

Finally, you understand what I mean. Thank God, you answered yourself, haha  

Nothing change, no exploit/vulnerability anyway.

Yes, the puzzles are completely random.

it's not random. it's a Hierarchical Deterministic Wallet cutted on MSB and with first byte adjusted for be into the range. Maybe only the first byte it's choosed randomly (or maybe not  )

Finally, you understand what I mean. Thank God, you answered yourself, haha  

Nothing change, no exploit/vulnerability anyway.

Guys, just stop feeding into his delusion. He found one, something that no other mathematician or cryptographer could ever do in the last decade.

Just leave it at that and move on.

I am not here to make you believe me. I am just saying what I have found out over these months. If you do not understand that these puzzles have no weakness, then that is your problem.  

You're wrong, there's a vulnerability and I've already found it.  

Yes, the puzzles are completely random.

it's not random. it's a Hierarchical Deterministic Wallet cutted on MSB and with first byte adjusted for be into the range. Maybe only the first byte it's choosed randomly (or maybe not  )

Finally, you understand what I mean. Thank God, you answered yourself, haha  

Nothing change, no exploit/vulnerability anyway.

Guys, just stop feeding into his delusion. He found one, something that no other mathematician or cryptographer could ever do in the last decade.

Just leave it at that and move on.

I am not here to make you believe me. I am just saying what I have found out over these months. If you do not understand that these puzzles have no weakness, then that is your problem.