Skip to content
Bitcoin Address Prefix Finder
C++ Objective-C C Cuda Makefile
Branch: master
Clone or download
Type Name Latest commit message Commit time
Failed to load latest commit information.
GPU Linux fixes May 13, 2019
hash Linux fixes May 13, 2019
.gitignore Patch for VS2015 and CUDA8 Mar 30, 2019
Base58.cpp Code cleaning Mar 27, 2019
Base58.h Improved loading of large list Mar 20, 2019
Bech32.cpp Added support for BECH32 Apr 3, 2019
Bech32.h Added support for BECH32 Apr 3, 2019
Int.cpp Patch for VS2015 and CUDA8 Mar 30, 2019
Int.h Added partial private key search Apr 17, 2019
IntGroup.cpp Support for multiple prefixes Mar 8, 2019
IntGroup.h Performance increase Mar 6, 2019
IntMod.cpp Added partial private key search Apr 17, 2019
LICENSE.txt Added license Feb 20, 2019
Makefile Linux fixes May 13, 2019
Point.cpp Added GPL header Feb 24, 2019
Point.h Added GPL header Feb 24, 2019 Update May 16, 2019
Random.cpp Added GPL header Feb 24, 2019
Random.h Added GPL header Feb 24, 2019
SECP256K1.cpp Prevent from generating Segwit address with uncompreesed key May 20, 2019
SECP256k1.h Added wildcard search May 13, 2019
Timer.cpp Revert May 14, 2019
Timer.h Switch to CryptoSecureRNG May 13, 2019
Vanity.cpp Change blinking pointer/cursor position to end Jul 16, 2019
Vanity.h Added wildcard search May 13, 2019
VanitySearch.sln Performance increase Feb 23, 2019
VanitySearch.vcxproj Added wildcard search May 13, 2019
VanitySearch.vcxproj.filters Added wildcard search May 13, 2019
VanitySearchCUDA8.sln Patch for VS2015 and CUDA8 Mar 30, 2019
VanitySearchCUDA8.vcxproj Updated CUD8 project Apr 14, 2019
VanitySearchCUDA8.vcxproj.filters Updated CUD8 project Apr 14, 2019
Wildcard.cpp Added wildcard search May 13, 2019
Wildcard.h Added wildcard search May 13, 2019
main.cpp Prevent from generating Segwit address with uncompreesed key May 20, 2019


VanitySearch is a bitcoin address prefix finder. If you want to generate safe private keys, use the -s option to enter your passphrase which will be used for generating a base key as for BIP38 standard (VanitySeacrh.exe -s "My PassPhrase" 1MyPrefix). You can also use VanitySeacrh.exe -ps "My PassPhrase" which will add a crypto secure seed to your passphrase.
VanitySearch may not compute a good grid size for your GPU, so try different values using -g option in order to get the best performances. If you want to use GPUs and CPUs together, you may have best performances by keeping one CPU core for handling GPU(s)/CPU exchanges (use -t option to set the number of CPU threads).


  • Fixed size arithmetic
  • Fast Modular Inversion (Delayed Right Shift 62 bits)
  • SecpK1 Fast modular multiplication (2 steps folding 512bits to 256bits using 64 bits digits)
  • Use some properties of elliptic curve to generate more keys
  • SSE Secure Hash Algorithm SHA256 and RIPEMD160 (CPU)
  • Multi-GPU support
  • CUDA optimisation via inline PTX assembly
  • Seed protected by pbkdf2_hmac_sha512 (BIP38)
  • Support P2PKH, P2SH and BECH32 addresses
  • Support split-key vanity address

Discussion Thread

Disucussion about VanitySearch@bitcointalk


You can downlad latest release from

VanitySeacrh [-check] [-v] [-u] [-b] [-c] [-gpu] [-stop] [-i inputfile]
             [-gpuId gpuId1[,gpuId2,...]] [-g gridSize1[,gridSize2,...]]
             [-o outputfile] [-m maxFound] [-ps seed] [-s seed] [-t nbThread]
             [-nosse] [-r rekey] [-check] [-kp] [-sp startPubKey]
             [-rp privkey partialkeyfile] [prefix]

 prefix: prefix to search (Can contains wildcard '?' or '*')
 -v: Print version
 -u: Search uncompressed addresses
 -b: Search both uncompressed or compressed addresses
 -c: Case unsensitive search
 -gpu: Enable gpu calculation
 -stop: Stop when all prefixes are found
 -i inputfile: Get list of prefixes to search from specified file
 -o outputfile: Output results to the specified file
 -gpu gpuId1,gpuId2,...: List of GPU(s) to use, default is 0
 -g gridSize1,gridSize2,...: Specify GPU(s) kernel gridsize, default is 8*(MP number)
 -m: Specify maximun number of prefixes found by each kernel call
 -s seed: Specify a seed for the base key, default is random
 -ps seed: Specify a seed concatened with a crypto secure random seed
 -t threadNumber: Specify number of CPU thread, default is number of core
 -nosse: Disable SSE hash function
 -l: List cuda enabled devices
 -check: Check CPU and GPU kernel vs CPU
 -kp: Generate key pair
 -rp privkey partialkeyfile: Reconstruct final private key(s) from partial key(s) info.
 -sp startPubKey: Start the search with a pubKey (for private key splitting)
 -r rekey: Rekey interval in MegaKey, default is disabled

Exemple (Windows, Intel Core i7-4770 3.4GHz 8 multithreaded cores, GeForce GTX 1050 Ti):

C:\C++\VanitySearch\x64\Release>VanitySearch.exe -stop -gpu 1TryMe
VanitySearch v1.11
Difficulty: 15318045009
Search: 1TryMe [Compressed]
Start Wed Apr  3 08:47:08 2019
Base Key:87B1EC7916A180ACCF07CAAEFA7F6508F3898F61AF49C201D70DF1543CCBA572
Number of CPU thread: 7
GPU: GPU #0 GeForce GTX 1050 Ti (6x128 cores) Grid(48x128)
245.830 MK/s (GPU 226.348 MK/s) (2^30.87) [P 12.06%][50.00% in 00:00:35][0]
Pub Addr: 1TryMeTKr3tuJZYHMSNWdPZfhRRNYj3yE
Priv (WIF): p2pkh:L5NuSjQRARifQJbZ5RyLrQhbSz25jYxupnqqydnBdANeH3QNoUph
Priv (HEX): 0xF36DD1EEC2A9658E50B39B280D4002ED3A07C7B6C07B37B191973BDDFBF9E375
C:\C++\VanitySearch\x64\Release>VanitySearch.exe -stop -gpu 3MyCoin
VanitySearch v1.11
Difficulty: 15318045009
Search: 3MyCoin [Compressed]
Start Wed Apr  3 14:52:45 2019
Base Key:FAF4F856077398AE087372110BF47A1A713C8F94B19CDD962D240B6A853CAD8B
Number of CPU thread: 7
GPU: GPU #0 GeForce GTX 1050 Ti (6x128 cores) Grid(48x128)
124.232 MK/s (GPU 115.601 MK/s) (2^33.18) [P 47.02%][50.00% in 00:00:07][0]
Pub Addr: 3MyCoinoA167kmgPprAidSvv5NoM3Nh6N3
Priv (WIF): p2wpkh-p2sh:L2qvghanHHov914THEzDMTpAyoRmxo7Rh85FLE9oKwYUrycWqudp
Priv (HEX): 0xA7D14FBF43696CA0B3DBFFD0AB7C9ED740FE338B2B856E09F2E681543A444D58
C:\C++\VanitySearch\x64\Release>VanitySearch.exe -stop -gpu bc1quantum
VanitySearch v1.11
Difficulty: 1073741824
Search: bc1quantum [Compressed]
Start Wed Apr  3 15:01:15 2019
Base Key:B00FD8CDA85B11D4744C09E65C527D35E231D19084FBCA0BF2E48186F31936AE
Number of CPU thread: 7
GPU: GPU #0 GeForce GTX 1050 Ti (6x128 cores) Grid(48x128)
256.896 MK/s (GPU 226.482 MK/s) (2^28.94) [P 38.03%][50.00% in 00:00:00][0]
Pub Addr: bc1quantum898l8mx5pkvq2x250kkqsj7enpx3u4yt
Priv (WIF): p2wpkh:L37xBVcFGeAZ9Tii7igqXBWmfiBhiwwiKQmchNXPV2LNREXQDLCp
Priv (HEX): 0xB00FD8CDA85B11D4744C09E65C527D35E2B1D19095CFCA0BF2E48186F31979C2

Generate a vanity address for a third party using split-key

It is possible to generate a vanity address for a third party in a safe manner using split-key.
For instance, Alice wants a nice prefix but does not have CPU power. Bob has the requested CPU power but cannot know the private key of Alice, Alice has to use a split-key.

Step 1

Alice generates a key pair on her computer then send the generated public key and the wanted prefix to Bob. It can be done by email, nothing is secret. Nevertheless, Alice has to keep safely the private key and not expose it.

VanitySearch.exe -s "AliceSeed" -kp
Priv : L4U2Ca2wyo721n7j9nXM9oUWLzCj19nKtLeJuTXZP3AohW9wVgrH
Pub  : 03FC71AE1E88F143E8B05326FC9A83F4DAB93EA88FFEACD37465ED843FCC75AA81

Note: The key pair is a standard SecpK1 key pair and can be generated with a third party software.

Step 2

Bob runs VanitySearch using the Alice's public key and the wanted prefix.

VanitySearch.exe -sp 03FC71AE1E88F143E8B05326FC9A83F4DAB93EA88FFEACD37465ED843FCC75AA81 -gpu -stop -o keyinfo.txt 1ALice

It generates a keyinfo.txt file containing the partial private key.

Pub Addr: 1ALicegohz9YgrLLa4ADCmam7X2Zr6xJZx
PartialPriv: L2hbovuDd8nG4nxjDq1yd5qDsSQiG8xFsAFbHMcThqfjSP6WLg89

Bob sends back this file to Alice. It can also be done by email. The partial private key does not allow anyone to guess the final Alice's private key.

Step 3

Alice can then reconstructs the final private key using her private key (the one generated in step 1) and the keyinfo.txt from Bob.

VanitySearch.exe -rp L4U2Ca2wyo721n7j9nXM9oUWLzCj19nKtLeJuTXZP3AohW9wVgrH keyinfo.txt

Pub Addr: 1ALicegohz9YgrLLa4ADCmam7X2Zr6xJZx
Priv (WIF): p2pkh:L1NHFgT826hYNpNN2qd85S7F7cyZTEJ4QQeEinsCFzknt3nj9gqg
Priv (HEX): 0x7BC226A19A1E9770D3B0584FF2CF89E5D43F0DC19076A7DE1943F284DA3FB2D0

How it works

Basically the -sp (start public key) adds the specified starting public key (let's call it Q) to the starting keys of each threads. That means that when you search (using -sp), you do not search for addr(k.G) but for addr(kpart.G+Q) where k is the private key in the first case and kpart the "partial private key" in the second case. G is the SecpK1 generator point.
Then the requester can reconstruct the final private key by doing kpart+ksecret (mod n) where kpart is the partial private key found by the searcher and ksecret is the private key of Q (Q=ksecret.G). This is the purpose of the -rp option.
The searcher has found a match for addr(kpart.G+ksecret.G) without knowing ksecret so the requester has the wanted address addr(kpart.G+Q) and the corresponding private key kpart+ksecret (mod n). The searcher is not able to guess this final private key because he doesn't know ksecret (he knows only Q).

Note: This explanation is simplified, it does not take care of symmetry and endomorphism optimizations but the idea is the same.

Trying to attack a list of addresses

The bitcoin address (P2PKH) consists of a hash160 (displayed in Base58 format) which means that there are 2160 possible addresses. A secure hash function can be seen as a pseudo number generator, it transforms a given message in a random number. In this case, a number (uniformaly distributed) in the range [0,2160]. So, the probability to hit a particular number after n tries is 1-(1-1/2160)n. We perform n Bernoulli trials statistically independent.
If we have a list of m distinct addresses (m<=2160), the search space is then reduced to 2160/m, the probability to find a collision after 1 try becomes m/2160 and the probability to find a collision after n tries becomes 1-(1-m/2160)n.
An example:
We have a hardware capable of generating 1GKey/s and we have an input list of 106 addresses, the following table shows the probability of finding a collision after a certain amount of time:

Time Probability
1 second 6.8e-34
1 minute 4e-32
1 hour 2.4e-30
1 day 5.9e-29
1 year 2.1e-26
10 years 2.1e-25
1000 years 2.1e-23
Age of earth 8.64e-17
Age of universe 2.8e-16 (much less than winning at the lottery)

Calculation has been done using this online high precision calculator

As you can see, even with a competitive hardware, it is very unlikely that you find a collision. Birthday paradox doesn't apply in this context, it works only if we know already the public key (not the address, the hash of the public key) we want to find. This program doesn't look for collisions between public keys. It searchs only for collisions with addresses with a certain prefix.



Intall CUDA SDK and open VanitySearch.sln in Visual C++ 2017.
You may need to reset your Windows SDK version in project properties.
In Build->Configuration Manager, select the Release configuration.
Build and enjoy.

Note: The current relase has been compiled with CUDA SDK 10.0, if you have a different release of the CUDA SDK, you may need to update CUDA SDK paths in VanitySearch.vcxproj using a text editor. The current nvcc option are set up to architecture starting at 3.0 capability, for older hardware, add the desired compute capabilities to the list in properties, CUDA C/C++, Device, Code Generation.


Intall CUDA SDK.
Depenging on the CUDA SDK version and on your Linux distribution you may need to install an older g++ (just for the CUDA SDK).
Edit the makefile and set up the good CUDA SDK path and appropriate compiler for nvcc.

CUDA       = /usr/local/cuda-8.0
CXXCUDA    = /usr/bin/g++-4.8

You can enter a list of architectrure (refer to nvcc documentation) if you have several GPU with different architecture. Compute capability 2.0 (Fermi) is deprecated for recent CUDA SDK. VanitySearch need to be compiled and linked with a recent gcc (>=7). The current release has been compiled with gcc 7.3.0.
Go to the VanitySearch directory. ccap is the desired compute capability.

$ g++ -v
gcc version 7.3.0 (Ubuntu 7.3.0-27ubuntu1~18.04)
$ make all (for build without CUDA support)
$ make gpu=1 ccap=20 all

Runnig VanitySearch (Intel(R) Xeon(R) CPU, 8 cores, @ 2.93GHz, Quadro 600 (x2))

$export LD_LIBRARY_PATH=/usr/local/cuda-8.0/lib64
pons@linpons:~/VanitySearch$ ./VanitySearch -t 7 -gpu -gpuId 0,1 1TryMe
VanitySearch v1.10
Difficulty: 15318045009
Search: 1TryMe [Compressed]
Start Wed Mar 27 10:26:43 2019
Base Key:C6718D8E50C1A5877DE3E52021C116F7598826873C61496BDB7CAD668CE3DCE5
Number of CPU thread: 7
GPU: GPU #1 Quadro 600 (2x48 cores) Grid(16x128)
GPU: GPU #0 Quadro 600 (2x48 cores) Grid(16x128)
40.284 MK/s (GPU 27.520 MK/s) (2^31.84) [P 22.24%][50.00% in 00:02:47][0]  
Pub Addr: 1TryMeERTZK7RCTemSJB5SNb2WcKSx45p
Priv (WIF): Ky9bMLDpb9o5rBwHtLaidREyA6NzLFkWJ19QjPDe2XDYJdmdUsRk
Priv (HEX): 0x398E7271AF3E5A78821C1ADFDE3EE90760A6B65F72D856CFE455B1264350BCE8


VanitySearch is licensed under GPLv3.

You can’t perform that action at this time.