📜 New blog: Real-World Assets on TensorCash — the ticker is the easy part.
A blockchain moves a unit pseudonymously, efficiently, across borders. What it cannot do by itself is say what the unit MEANS. If a token represents cash, reserves, a fund or a share, someone must define the claim and sign the terms — and the asset keeps living after issuance: distributions, votes, amendments.
How TensorCash joins the two halves:
• Consensus-enforced issuance parameters — denomination, hard live-supply cap, mint/burn policy, script and KYC rules. Visible in the explorer, enforced by every node.
• The legal text travels with the asset, anchored by its on-chain control output. Public mode for broad-disclosure instruments; holder-only mode keeps the document encrypted on-chain while its commitments stay verifiable by everyone. Worked examples: a live private share-administration issuance on mainnet, and UKLTD — a public educational model — on testnet.
• Text/witness split — the canonical text is frozen and hashed; qualified e-signatures attest to that exact hash in a separate witness bundle. Holders accept the designated clauses with an on-chain 0x40 record.
• Full lifecycle in the desktop wallet — mint, burn, compliance, distributions, holder votes, and option/CFD series referencing the asset itself.
RWAs as first-class citizens: they can be financed, exchanged, and have derivatives written on them — where their governing documents and applicable law permit.
https://tensorcash.org/blog/rwa-the-ticker-is-the-easy-part/
A blockchain moves a unit pseudonymously, efficiently, across borders. What it cannot do by itself is say what the unit MEANS. If a token represents cash, reserves, a fund or a share, someone must define the claim and sign the terms — and the asset keeps living after issuance: distributions, votes, amendments.
How TensorCash joins the two halves:
• Consensus-enforced issuance parameters — denomination, hard live-supply cap, mint/burn policy, script and KYC rules. Visible in the explorer, enforced by every node.
• The legal text travels with the asset, anchored by its on-chain control output. Public mode for broad-disclosure instruments; holder-only mode keeps the document encrypted on-chain while its commitments stay verifiable by everyone. Worked examples: a live private share-administration issuance on mainnet, and UKLTD — a public educational model — on testnet.
• Text/witness split — the canonical text is frozen and hashed; qualified e-signatures attest to that exact hash in a separate witness bundle. Holders accept the designated clauses with an on-chain 0x40 record.
• Full lifecycle in the desktop wallet — mint, burn, compliance, distributions, holder votes, and option/CFD series referencing the asset itself.
RWAs as first-class citizens: they can be financed, exchanged, and have derivatives written on them — where their governing documents and applicable law permit.
https://tensorcash.org/blog/rwa-the-ticker-is-the-easy-part/
TensorCash
Real-World Assets on TensorCash: the ticker is the easy part — TensorCash
A ticker is a label, not a legal relationship. How TensorCash joins consensus-enforced asset rules with a versioned legal document that travels with the asset — illustrated by a live private share-administration issuance on mainnet and UKLTD, the public educational…
It is the weekend—a good time to take a step back and look at what we have accomplished.
Pick a prompt. Any prompt you want. A real model responds. The response is perfectly natural and no different from a normal model response. There is no separate mining workload, but that inference can mine a block.
It is easy to take this for granted because it looks so simple and has run so smoothly. But no one had done it before: a normal model response and a mined block, produced by the same inference.
The same proof that mines TensorCash can be used as a receipt to sell the inference in the market. AI inference is set to be one of the defining markets of the next decade, and TensorCash gives independent GPU operators a chance to participate as providers—not spectators.
Newer and larger models are coming. The protocol path is already in the code; this is not waiting on a new engineering build.
Many of you are looking at difficulty and production cost today. We respect that. GPUs cost money, and electricity bills arrive at the end of the month.
But think about what difficulty looks like when a neocloud adopts the protocol across its inference workloads. Or when a Bitcoin miner converting sites into GPU datacenters sees this as the pivot. Or when a compute-constrained model provider taps into the decentralized network.
That is the point: this is not a milestone conditional on a future engineering build or development. It can happen tomorrow.
Pick a prompt. Any prompt you want. A real model responds. The response is perfectly natural and no different from a normal model response. There is no separate mining workload, but that inference can mine a block.
It is easy to take this for granted because it looks so simple and has run so smoothly. But no one had done it before: a normal model response and a mined block, produced by the same inference.
The same proof that mines TensorCash can be used as a receipt to sell the inference in the market. AI inference is set to be one of the defining markets of the next decade, and TensorCash gives independent GPU operators a chance to participate as providers—not spectators.
Newer and larger models are coming. The protocol path is already in the code; this is not waiting on a new engineering build.
Many of you are looking at difficulty and production cost today. We respect that. GPUs cost money, and electricity bills arrive at the end of the month.
But think about what difficulty looks like when a neocloud adopts the protocol across its inference workloads. Or when a Bitcoin miner converting sites into GPU datacenters sees this as the pivot. Or when a compute-constrained model provider taps into the decentralized network.
That is the point: this is not a milestone conditional on a future engineering build or development. It can happen tomorrow.
❤5🔥1
What sets TensorCash apart?
Everyone wants to do AI + crypto. It is cool. It is hyped. Every project now claims a Proof of Something.
The useful question is simple: what does each protocol actually prove?
TensorCash puts real model inference in the block and lets that inference be sold on the open market without trusting the provider. The proof both mines the block and serves as the inference receipt. Buyers verify execution, then compare providers on price and latency. The response is the work, and one inference cannot mine two blocks.
Now group the alternatives by what they claim to prove:
• Proof of Model — Keryx. It proves model weights are resident in GPU memory. Residency is not inference, and Keryx needs a separate mechanism for answer correctness. What stops one GPU claiming the same resident model across multiple reward networks?
• Proof of GPU, capacity and attestation — io.net, Aethir, Phala. These prove hardware, availability or an attested stack. What proves that capacity is not double-booked across networks? Attestation trusts the vendor chain—not the inference.
• Proof of MatMul — BTX, Pearl. The proof rewards matrix work rather than the customer response. If mining while inferring makes both slower, rational operators specialize: mine, infer, or split the hardware. The efficient miner strips away inference. What remains is a GPU puzzle—not an inference receipt. Nockchain points to the same arbitrary-MatMul cuPoW as a future AI route, so the same critique applies.
• Proof of Transformer and benchmarked capacity — Gonka. It benchmarks a randomized transformer Sprint, awards compute power, then allocates inference. The benchmark is not the service. TensorCash makes the service itself proof-of-work.
• Prover–verifier networks — Bittensor. Miners produce subnet work; validators score it. That proves who validators rewarded, not which model executed. Its documentation describes weight copying: predicting consensus can replace evaluation.
• Proof of Logits — Ambient. “Hardware agnostic” needs a specification: cross-GPU calibration, false-accept and false-reject rates, with boundaries that tolerate honest variation but catch cheaper execution.
TensorCash already did the verification work. Every node runs deterministic code with calibrated, model-specific statistical boundaries. Honest GPU variation passes. The live network distinguished bf16 from fp8 on the same model; the fp8 blocks received zero work.
The published evaluation measured single-digit proof-emission overhead. Production mining software improved logging performance and brought measured overhead to zero; the earlier residual was a logging artefact.
That full combination is the distinction: real inference as block work, one-use work, deterministic verification, and a receipt that lets the inference be sold on the open market without provider trust. Unknown operators and datacenters compete on price and latency.
If another live chain does all of this, send the block, proof, verifier and end-to-end throughput numbers. We are here to talk.
Read the proof, not the pitch:
https://tensorcash.org/blog/what-sets-tensorcash-apart/
Everyone wants to do AI + crypto. It is cool. It is hyped. Every project now claims a Proof of Something.
The useful question is simple: what does each protocol actually prove?
TensorCash puts real model inference in the block and lets that inference be sold on the open market without trusting the provider. The proof both mines the block and serves as the inference receipt. Buyers verify execution, then compare providers on price and latency. The response is the work, and one inference cannot mine two blocks.
Now group the alternatives by what they claim to prove:
• Proof of Model — Keryx. It proves model weights are resident in GPU memory. Residency is not inference, and Keryx needs a separate mechanism for answer correctness. What stops one GPU claiming the same resident model across multiple reward networks?
• Proof of GPU, capacity and attestation — io.net, Aethir, Phala. These prove hardware, availability or an attested stack. What proves that capacity is not double-booked across networks? Attestation trusts the vendor chain—not the inference.
• Proof of MatMul — BTX, Pearl. The proof rewards matrix work rather than the customer response. If mining while inferring makes both slower, rational operators specialize: mine, infer, or split the hardware. The efficient miner strips away inference. What remains is a GPU puzzle—not an inference receipt. Nockchain points to the same arbitrary-MatMul cuPoW as a future AI route, so the same critique applies.
• Proof of Transformer and benchmarked capacity — Gonka. It benchmarks a randomized transformer Sprint, awards compute power, then allocates inference. The benchmark is not the service. TensorCash makes the service itself proof-of-work.
• Prover–verifier networks — Bittensor. Miners produce subnet work; validators score it. That proves who validators rewarded, not which model executed. Its documentation describes weight copying: predicting consensus can replace evaluation.
• Proof of Logits — Ambient. “Hardware agnostic” needs a specification: cross-GPU calibration, false-accept and false-reject rates, with boundaries that tolerate honest variation but catch cheaper execution.
TensorCash already did the verification work. Every node runs deterministic code with calibrated, model-specific statistical boundaries. Honest GPU variation passes. The live network distinguished bf16 from fp8 on the same model; the fp8 blocks received zero work.
The published evaluation measured single-digit proof-emission overhead. Production mining software improved logging performance and brought measured overhead to zero; the earlier residual was a logging artefact.
That full combination is the distinction: real inference as block work, one-use work, deterministic verification, and a receipt that lets the inference be sold on the open market without provider trust. Unknown operators and datacenters compete on price and latency.
If another live chain does all of this, send the block, proof, verifier and end-to-end throughput numbers. We are here to talk.
Read the proof, not the pitch:
https://tensorcash.org/blog/what-sets-tensorcash-apart/
TensorCash
What sets TensorCash apart — TensorCash
TensorCash makes real inference the block work, then turns the mining proof into the receipt that lets the inference be sold on the open market without trusting the provider.
👍2🔥2
🔗 The reply that mines the chain
Protocols call themselves ecosystems long before they have one. The test is whether people you have never met build things you did not commission — then put them in front of real users.
That is beginning to happen on TensorCash:
🤖 Bob Labs — chat.tensorcash.boblabs.eu: a full chat + agent-orchestration platform (agents, multi-agent labs, tools, MCP, memory, cron, and API usage prepaid in TSC) running inside a mining pool. The inference behind each reply becomes mining work. No card or identity is requested at sign-up.
🏢 AskTilly — an enterprise-grade chat and agent platform: passkey-first and end-to-end confidential, with an API and a self-custody WebWallet (client-side signing, passkey PRF unlock, SegWit, Taproot and post-quantum addresses, and signed desktop builds).
⛏️ Three public pools: luckypool.io, tsc.suprnova.cc and Bob Labs. Three pools do not guarantee decentralisation, but they give miners something essential: a choice.
Running something on TensorCash? Add it to the community directory — one PR.
https://tensorcash.org/blog/third-party-ecosystem/
Protocols call themselves ecosystems long before they have one. The test is whether people you have never met build things you did not commission — then put them in front of real users.
That is beginning to happen on TensorCash:
🤖 Bob Labs — chat.tensorcash.boblabs.eu: a full chat + agent-orchestration platform (agents, multi-agent labs, tools, MCP, memory, cron, and API usage prepaid in TSC) running inside a mining pool. The inference behind each reply becomes mining work. No card or identity is requested at sign-up.
🏢 AskTilly — an enterprise-grade chat and agent platform: passkey-first and end-to-end confidential, with an API and a self-custody WebWallet (client-side signing, passkey PRF unlock, SegWit, Taproot and post-quantum addresses, and signed desktop builds).
⛏️ Three public pools: luckypool.io, tsc.suprnova.cc and Bob Labs. Three pools do not guarantee decentralisation, but they give miners something essential: a choice.
Running something on TensorCash? Add it to the community directory — one PR.
https://tensorcash.org/blog/third-party-ecosystem/
TensorCash
The reply that mines the chain: inside TensorCash's third-party ecosystem — TensorCash
A chat platform where AI replies become mining work, a confidential agent platform with browser self-custody, and three public pools — all built by independent operators.
👍4🔥1
✨ **TensorCash Magic #1: mining inference is normal inference.**
Ask an AI a question. It answers exactly as it normally would.
Same model. Same next-word probabilities. Same answer quality.
But that useful inference is also doing the work of mining.
There is no dummy calculation running beside it. No weaker “mining model.” No trade between a useful answer and network security.
So where is the magic?
Inside every AI answer is a hidden coin flip. At each step, the model proposes possible next tokens — *Light*, *Sunlight*, *Rainbows* — with different odds. A sampler draws a number and chooses one. Then the chosen word becomes part of the context, and the model does it again.
A normal chatbot gets that draw from a private RNG. TensorCash derives it from the chain and the answer so far.
The model never changes. The odds never change. The quality never changes. But the full sequential inference path is now bound to the chain, replayable and verifiable.
**Normal inference goes in. Verifiable mining work comes out.**
Watch three answers grow token by token: https://tensorcash.org/blog/chain-derived-sampling/
https://tensorcash.org/blog/chain-derived-sampling/
Ask an AI a question. It answers exactly as it normally would.
Same model. Same next-word probabilities. Same answer quality.
But that useful inference is also doing the work of mining.
There is no dummy calculation running beside it. No weaker “mining model.” No trade between a useful answer and network security.
So where is the magic?
Inside every AI answer is a hidden coin flip. At each step, the model proposes possible next tokens — *Light*, *Sunlight*, *Rainbows* — with different odds. A sampler draws a number and chooses one. Then the chosen word becomes part of the context, and the model does it again.
A normal chatbot gets that draw from a private RNG. TensorCash derives it from the chain and the answer so far.
The model never changes. The odds never change. The quality never changes. But the full sequential inference path is now bound to the chain, replayable and verifiable.
**Normal inference goes in. Verifiable mining work comes out.**
Watch three answers grow token by token: https://tensorcash.org/blog/chain-derived-sampling/
https://tensorcash.org/blog/chain-derived-sampling/
TensorCash
TensorCash magic: mining inference is the same as normal inference — TensorCash
TensorCash mining runs the same autoregressive generation as normal LLM inference. The model, top-50 probabilities and output quality stay unchanged; only the sampler's pseudorandom source is chain-derived.
👍3
Welcome to the Bitcoin genesis era. At block 22,165, the TensorCash subsidy moves from 92.664 to 55.5984 TSC — close to Bitcoin's original 50 BTC subsidy. Same ~21M destination; a different road there. Height, not date, decides it.
https://tensorcash.org/blog/welcome-to-the-bitcoin-genesis-era/
https://tensorcash.org/blog/welcome-to-the-bitcoin-genesis-era/
🔥3😁1
People often assume a useful-proof-of-work chain has to choose one AI model forever. TensorCash does not.
A model joins through an on-chain registration: a 5 TSC sponsor deposit, a 100-block window and at least 50 supporting commits in valid blocks. That makes admission a hashpower-backed consensus decision—not an admin switch.
Every registered model then competes for the same next block. nBits provides the network-wide base target; nAdjBits gives each model an adjusted lane based on its registered difficulty. Miners choose among those lanes using the economics of block rewards, inference demand and operating cost.
And no extra validator class is needed. The statistical tests are protocol-defined; nobody scores whether an answer “looks intelligent.”
One chain. N models. Quick explainer:
https://tensorcash.org/blog/how-tensorcash-runs-multiple-models/
A model joins through an on-chain registration: a 5 TSC sponsor deposit, a 100-block window and at least 50 supporting commits in valid blocks. That makes admission a hashpower-backed consensus decision—not an admin switch.
Every registered model then competes for the same next block. nBits provides the network-wide base target; nAdjBits gives each model an adjusted lane based on its registered difficulty. Miners choose among those lanes using the economics of block rewards, inference demand and operating cost.
And no extra validator class is needed. The statistical tests are protocol-defined; nobody scores whether an answer “looks intelligent.”
One chain. N models. Quick explainer:
https://tensorcash.org/blog/how-tensorcash-runs-multiple-models/
👍1
TensorCash is now listed on two exchanges.
Every coin in circulation was mined by real model inference — no premine, no allocation. Until now, moving between that work and the wider world mostly meant OTC or on-chain trades. This makes the path wider: miners who earn coins by serving inference and users who hold or spend them now have more ways in and out.
A stepping stone, not a destination — the goal remains an open compute market where inference is sold with a verifiable receipt. But stepping stones are worth celebrating.
And as ever: an exchange account is not a wallet. Your keys, your coins.
Every coin in circulation was mined by real model inference — no premine, no allocation. Until now, moving between that work and the wider world mostly meant OTC or on-chain trades. This makes the path wider: miners who earn coins by serving inference and users who hold or spend them now have more ways in and out.
A stepping stone, not a destination — the goal remains an open compute market where inference is sold with a verifiable receipt. But stepping stones are worth celebrating.
And as ever: an exchange account is not a wallet. Your keys, your coins.
👍3
**TensorCash v1.1.1 — verifier release: upgrade ahead of Qwen3.8**
Qwen3.5-family models register with a fixed one-window sampling profile: every EOS/stop token is suppressed for all 256 proof steps (min_tokens=256). The inference engine applies that before the PoW snapshot is captured, but replay was scoring raw logits — so honest end-of-segment windows could read as RED. v1.1.1 replays the same suppression before snapshot/CDF/rank/Mahalanobis, so honest proofs verify GREEN and every other model family stays byte-identical.
Also in this release:
• Mask-aware MCA context scaling with a bounded long-context multiplier
• MCA coverage for Gated DeltaNet sequence mixers
• Padded-vocab -inf snapshot guard (multi-GPU topologies)
• transformers 5.14.1 / huggingface-hub 1.5.0
If you verify blocks — pools, standalone verifiers, pool-verify-pack users — upgrade before Qwen3.8-family registration: older verifiers will raise false REDs on honest Qwen3.8 proofs at segment ends.
https://git.tensorcash.org/tensorcash/tensorcash/releases/tag/v1.1.1
Qwen3.5-family models register with a fixed one-window sampling profile: every EOS/stop token is suppressed for all 256 proof steps (min_tokens=256). The inference engine applies that before the PoW snapshot is captured, but replay was scoring raw logits — so honest end-of-segment windows could read as RED. v1.1.1 replays the same suppression before snapshot/CDF/rank/Mahalanobis, so honest proofs verify GREEN and every other model family stays byte-identical.
Also in this release:
• Mask-aware MCA context scaling with a bounded long-context multiplier
• MCA coverage for Gated DeltaNet sequence mixers
• Padded-vocab -inf snapshot guard (multi-GPU topologies)
• transformers 5.14.1 / huggingface-hub 1.5.0
If you verify blocks — pools, standalone verifiers, pool-verify-pack users — upgrade before Qwen3.8-family registration: older verifiers will raise false REDs on honest Qwen3.8 proofs at segment ends.
https://git.tensorcash.org/tensorcash/tensorcash/releases/tag/v1.1.1
TensorCash Git
v1.1.1 - tensorcash/tensorcash
tensorcash - Useful proof-of-work for verifiable AI — the TensorCash umbrella monorepo: miner integration, OSS proof-of-inference verifier, and core services for a Bitcoin-derived layer-1 secured by open LLM inference.
👍1
**Decentralisation Day.**
Around 850 distinct addresses received TSC mining rewards over the last four weeks. Five pools — Aria, Boblabs, LuckyPool, Suprnova, Tiger — integrated quickly, patiently and transparently, and are offering miners a choice. Roughly a fifth of all mined TSC now sits at addresses that never mined it. The project is increasingly in the community's hands.
The people who wrote the code and stabilised the network are not going anywhere — and a receipt follows. Block 24,000 will provide the public seed for drawing four early coinbase addresses. They will sign alongside the genesis key, and the statement hash will then be committed on-chain. Details in the post.
Don't trust — verify.
https://tensorcash.org/blog/decentralisation-day/
Around 850 distinct addresses received TSC mining rewards over the last four weeks. Five pools — Aria, Boblabs, LuckyPool, Suprnova, Tiger — integrated quickly, patiently and transparently, and are offering miners a choice. Roughly a fifth of all mined TSC now sits at addresses that never mined it. The project is increasingly in the community's hands.
The people who wrote the code and stabilised the network are not going anywhere — and a receipt follows. Block 24,000 will provide the public seed for drawing four early coinbase addresses. They will sign alongside the genesis key, and the statement hash will then be committed on-chain. Details in the post.
Don't trust — verify.
https://tensorcash.org/blog/decentralisation-day/
TensorCash v1.1.2 — mainnet upgrade required for block producers
Activation: mainnet height 24150 · testnet unchanged.
Deploy v1.1.2 before height 24150; consensus-validating nodes should also upgrade. It tightens proof-of-inference validation — after activation, blocks that don't meet the updated requirements are rejected by upgraded nodes. Forward-only; history unaffected (~0.25% of past blocks would not qualify, measured retrospectively).
Also: stale-verdict rollback fix, batched work-unit issuance, numeric RPC for batch units, mining-RPC guard.
Required — TensorCash 0f61b409fb40e41b1f715933e0c86a5be6b29741, bcore 10f6ade2374a82db73325dcee7edf0477c686ec1
An outdated producer may create a rejected block — please upgrade before height 24150.
https://git.tensorcash.org/tensorcash/tensorcash/releases/tag/v1.1.2
Activation: mainnet height 24150 · testnet unchanged.
Deploy v1.1.2 before height 24150; consensus-validating nodes should also upgrade. It tightens proof-of-inference validation — after activation, blocks that don't meet the updated requirements are rejected by upgraded nodes. Forward-only; history unaffected (~0.25% of past blocks would not qualify, measured retrospectively).
Also: stale-verdict rollback fix, batched work-unit issuance, numeric RPC for batch units, mining-RPC guard.
Required — TensorCash 0f61b409fb40e41b1f715933e0c86a5be6b29741, bcore 10f6ade2374a82db73325dcee7edf0477c686ec1
An outdated producer may create a rejected block — please upgrade before height 24150.
https://git.tensorcash.org/tensorcash/tensorcash/releases/tag/v1.1.2
TensorCash v1.1.2a — patch release
v1.1.2a fixes a bug in v1.1.2. If you've already updated to v1.1.2, please upgrade to v1.1.2a. Activation is unchanged at mainnet height 24150 — be on v1.1.2a before then.
Required versions:
TensorCash: 9bdeb44d76ef93f24f42b88c78cf1ed3402594f4
bcore: 8483a19095cef56203bd21d8e8aebbba9d98bb6f
https://git.tensorcash.org/tensorcash/tensorcash/releases/tag/v1.1.2a
v1.1.2a fixes a bug in v1.1.2. If you've already updated to v1.1.2, please upgrade to v1.1.2a. Activation is unchanged at mainnet height 24150 — be on v1.1.2a before then.
Required versions:
TensorCash: 9bdeb44d76ef93f24f42b88c78cf1ed3402594f4
bcore: 8483a19095cef56203bd21d8e8aebbba9d98bb6f
https://git.tensorcash.org/tensorcash/tensorcash/releases/tag/v1.1.2a
TensorCash Git
TensorCash v1.1.2a - tensorcash/tensorcash
# TensorCash v1.1.2a
## Mainnet upgrade for block producers
v1.1.2a supersedes the withdrawn v1.1.2 release and corrects its production proof-evidence validation.
- **Activation:** mainnet height **24150**
- **Testnet:** unchanged
- **Scope:** bcore update…
## Mainnet upgrade for block producers
v1.1.2a supersedes the withdrawn v1.1.2 release and corrects its production proof-evidence validation.
- **Activation:** mainnet height **24150**
- **Testnet:** unchanged
- **Scope:** bcore update…
**Block 24,000 is in — Decentralisation Day, signed.**
Its hash 46af892e8ac57e1a8a1c297e3e980565f4a83c759ce086be5f856f3058601706 drew blocks 1039, 683, 1439 and 246 from the first 2,000. Recompute it yourself: read the hash in 8-hex-digit chunks, take each mod 2000 + 1, keep the first four distinct heights. Those four coinbase addresses and the genesis key — whose public half has been on-chain since block zero — have signed the statement (BIP322 for the four, legacy signmessage for genesis), and the statement's SHA-256 is committed on-chain via OP_RETURN in block 24,235, txid 99054df7dbf0825cf2ee65bad0aa12a08e7e55648713914f3973e86147cd9167. Statement, signatures and draw are on the post.
Don't trust — verify.
https://tensorcash.org/blog/decentralisation-day/
Its hash 46af892e8ac57e1a8a1c297e3e980565f4a83c759ce086be5f856f3058601706 drew blocks 1039, 683, 1439 and 246 from the first 2,000. Recompute it yourself: read the hash in 8-hex-digit chunks, take each mod 2000 + 1, keep the first four distinct heights. Those four coinbase addresses and the genesis key — whose public half has been on-chain since block zero — have signed the statement (BIP322 for the four, legacy signmessage for genesis), and the statement's SHA-256 is committed on-chain via OP_RETURN in block 24,235, txid 99054df7dbf0825cf2ee65bad0aa12a08e7e55648713914f3973e86147cd9167. Statement, signatures and draw are on the post.
Don't trust — verify.
https://tensorcash.org/blog/decentralisation-day/
❤2🔥1
TensorCash v1.1.3 — central VDF beacon
The VDF makes the network more secure, and it doesn't have to cost anyone anything. We now offer an optional central VDF beacon: a public endpoint serving a verifiable (tick, proof) pair for the current tip. Your miner verifies it locally before use — nothing on trust — and v1.1.3 integrates it automatically, taking the higher of the beacon's tick and its own. Pools and solo miners are welcome to integrate it at their convenience.
Endpoint: https://mempool.tensorcash.org/api/v1/vdf
Chainwork amendments are being considered for a later release. Happy mining.
https://git.tensorcash.org/tensorcash/tensorcash/releases/tag/v1.1.3
The VDF makes the network more secure, and it doesn't have to cost anyone anything. We now offer an optional central VDF beacon: a public endpoint serving a verifiable (tick, proof) pair for the current tip. Your miner verifies it locally before use — nothing on trust — and v1.1.3 integrates it automatically, taking the higher of the beacon's tick and its own. Pools and solo miners are welcome to integrate it at their convenience.
Endpoint: https://mempool.tensorcash.org/api/v1/vdf
Chainwork amendments are being considered for a later release. Happy mining.
https://git.tensorcash.org/tensorcash/tensorcash/releases/tag/v1.1.3
TensorCash v1.1.4 is out — consensus tightening, activation at mainnet block 24,735 (≈ today 15:00 UTC at the current block rate).
From block 24,735 the network enforces a stricter validity rule on mining proofs. A block producer still on an older version risks producing an invalid block or building on an invalid block after activation.
Must upgrade before 24,735: block producers — pool operators and solo miners.
Can update at their leisure: pool participants (share-submitting miners); nothing changes for you at activation.
What's in it
• consensus: pad-masked prompt slots must be genuine padding (a left run of token 0); hiding prompt tokens from the model behind the pad mask is no longer valid (PadSlotZeroHeight = 24,735)
• verifier: minimum model-determined coherence on the prompt (prompt gibberish gate)
Get it
• release: https://git.tensorcash.org/tensorcash/tensorcash/releases/tag/v1.1.4
• image: docker.io/tensorcash/core-node:v1.1.4-amd64
From block 24,735 the network enforces a stricter validity rule on mining proofs. A block producer still on an older version risks producing an invalid block or building on an invalid block after activation.
Must upgrade before 24,735: block producers — pool operators and solo miners.
Can update at their leisure: pool participants (share-submitting miners); nothing changes for you at activation.
What's in it
• consensus: pad-masked prompt slots must be genuine padding (a left run of token 0); hiding prompt tokens from the model behind the pad mask is no longer valid (PadSlotZeroHeight = 24,735)
• verifier: minimum model-determined coherence on the prompt (prompt gibberish gate)
Get it
• release: https://git.tensorcash.org/tensorcash/tensorcash/releases/tag/v1.1.4
• image: docker.io/tensorcash/core-node:v1.1.4-amd64
TensorCash Git
v1.1.4 - tensorcash/tensorcash
tensorcash - Useful proof-of-work for verifiable AI — the TensorCash umbrella monorepo: miner integration, OSS proof-of-inference verifier, and core services for a Bitcoin-derived layer-1 secured by open LLM inference.
👍1
TensorCash v1.1.4a is out — a follow-up to v1.1.4.
Update your bcore node today, before block height 24870. Recommended for all node operators; mandatory for block producers.
Public pools and verification services are already updated. Miner images to follow within the next week; pools will advise.
We'll follow up with reminders and updates in due course. Thanks for your support.
Update your bcore node today, before block height 24870. Recommended for all node operators; mandatory for block producers.
Public pools and verification services are already updated. Miner images to follow within the next week; pools will advise.
We'll follow up with reminders and updates in due course. Thanks for your support.
📢 TensorCash StepBind (TIP-0003): the legacy grace period ends tonight, 23:59:59 UTC on 8 September.
StepBind activated at block 24,870 on 1 September. It binds every sampling step of a mining proof to the whole prompt, so evidence from one prompt cannot be reused for another. Since activation, proofs without the StepBind flag have been accepted under a grace rule. That grace ends tonight: after 23:59:59 UTC, blocks carrying legacy (flagless) proofs are invalid and every upgraded node rejects them. The cutoff is measured on chain median time, so it bites within about an hour after midnight UTC. Do not rely on that margin.
Where the network stands: of the 135 blocks mined in the last 24 hours, 71 still carry legacy proofs. The miners behind them stop earning at the cutoff unless they upgrade.
⚠️ What to do before midnight UTC:
• Node operators: run v1.1.4c. Docker: tensorcash/core-node:v1.1.4c-amd64 (pin the tag; latest is still an older build). A node on older code stalls at the first StepBind block it sees; v1.1.4c clears that on first start, no manual reconsiderblock needed.
• Miners: run StepBind-capable miner software. The v1.1.4c release ships llama-server and miner-proxy builds for Linux, plus Windows and Linux node packages. If you mine through a pool, check that your pool has switched its work templates to StepBind.
• Pools: your verifier must accept StepBind proofs and must stop crediting legacy shares at the cutoff.
Balances, wallets and transaction history are untouched; only proof validation changes.
Quick health check for any node: the height from getblockchaininfo should match the explorer at https://mempool.tensorcash.org. If your node is stuck at an earlier height, it is on pre-StepBind code.
https://git.tensorcash.org/tensorcash/tensorcash/releases/tag/v1.1.4c
StepBind activated at block 24,870 on 1 September. It binds every sampling step of a mining proof to the whole prompt, so evidence from one prompt cannot be reused for another. Since activation, proofs without the StepBind flag have been accepted under a grace rule. That grace ends tonight: after 23:59:59 UTC, blocks carrying legacy (flagless) proofs are invalid and every upgraded node rejects them. The cutoff is measured on chain median time, so it bites within about an hour after midnight UTC. Do not rely on that margin.
Where the network stands: of the 135 blocks mined in the last 24 hours, 71 still carry legacy proofs. The miners behind them stop earning at the cutoff unless they upgrade.
⚠️ What to do before midnight UTC:
• Node operators: run v1.1.4c. Docker: tensorcash/core-node:v1.1.4c-amd64 (pin the tag; latest is still an older build). A node on older code stalls at the first StepBind block it sees; v1.1.4c clears that on first start, no manual reconsiderblock needed.
• Miners: run StepBind-capable miner software. The v1.1.4c release ships llama-server and miner-proxy builds for Linux, plus Windows and Linux node packages. If you mine through a pool, check that your pool has switched its work templates to StepBind.
• Pools: your verifier must accept StepBind proofs and must stop crediting legacy shares at the cutoff.
Balances, wallets and transaction history are untouched; only proof validation changes.
Quick health check for any node: the height from getblockchaininfo should match the explorer at https://mempool.tensorcash.org. If your node is stuck at an earlier height, it is on pre-StepBind code.
https://git.tensorcash.org/tensorcash/tensorcash/releases/tag/v1.1.4c
👍1🔥1
Proof v4: the intelligence multiplier
Difficulty that reflects how the model generates.
We have been working on a substantial upgrade to consensus. PR #13 is open—please start reviewing: https://git.tensorcash.org/tensorcash/tensorcash/pulls/13
Proof v4 makes each block's difficulty reflect the sampling behaviour behind its proof. The aim is to keep ordinary inference close to its existing cost while making certain patterns of optimisation more expensive.
The method, research and charts:
https://tensorcash.org/blog/proof-v4-the-intelligence-multiplier/
The upgrade includes:
- Difficulty pricing by sampling profile: the intelligence multiplier.
- Gumbel-race sampling: one token-selection race using reproducible noise.
- Argon-bound sampling and result hashing: binding every draw and the final hash to a memory-hard output.
- Prompt and continuation reuse checks.
- Stricter verification: checks for sampling and the applicable activation rules.
Planned rollout (UTC):
Testnet activates overnight, 13–14 Sep, before the release.
1. Monday 14 Sep: we will cut the bcore + verifier release and publish all binaries.
2. Tuesday 15 Sep: pricing applies to both v3 and v4 proofs, with a one-off difficulty adjustment.
3. Sunday 20 Sep: the transition ends; v3 proofs are no longer accepted.
If more computational power shifts toward prompts engineered around the current proof boundaries, we will bring activation forward and announce the revised schedule.
Pools, miners and node operators: please review now and prepare to upgrade when the release is available. Share questions and findings here or on the PR.
Difficulty that reflects how the model generates.
We have been working on a substantial upgrade to consensus. PR #13 is open—please start reviewing: https://git.tensorcash.org/tensorcash/tensorcash/pulls/13
Proof v4 makes each block's difficulty reflect the sampling behaviour behind its proof. The aim is to keep ordinary inference close to its existing cost while making certain patterns of optimisation more expensive.
The method, research and charts:
https://tensorcash.org/blog/proof-v4-the-intelligence-multiplier/
The upgrade includes:
- Difficulty pricing by sampling profile: the intelligence multiplier.
- Gumbel-race sampling: one token-selection race using reproducible noise.
- Argon-bound sampling and result hashing: binding every draw and the final hash to a memory-hard output.
- Prompt and continuation reuse checks.
- Stricter verification: checks for sampling and the applicable activation rules.
Planned rollout (UTC):
Testnet activates overnight, 13–14 Sep, before the release.
1. Monday 14 Sep: we will cut the bcore + verifier release and publish all binaries.
2. Tuesday 15 Sep: pricing applies to both v3 and v4 proofs, with a one-off difficulty adjustment.
3. Sunday 20 Sep: the transition ends; v3 proofs are no longer accepted.
If more computational power shifts toward prompts engineered around the current proof boundaries, we will bring activation forward and announce the revised schedule.
Pools, miners and node operators: please review now and prepare to upgrade when the release is available. Share questions and findings here or on the PR.
👀2
TensorCash v1.2.0 — Proof v4 is here
Following yesterday's announcement, we are releasing v1.2.0: the intelligence multiplier and a substantial upgrade to consensus. Difficulty that reflects how the model generates.
Who should upgrade
• Pools, node operators and verification providers: upgrade today, 14 September, before activation at block 26,825.
• Pool miners: wait for your pool to confirm its upgrade, then follow its instructions to update your miner.
• Solo miners: upgrade as soon as possible.
Everyone must complete their upgrade before block 26,925 (Y), expected Tuesday 15 September. Pool miners should coordinate with their pool to meet this deadline.
Mainnet rollout
• 26,825 (X), expected overnight 14–15 September: Proof v4 activates. V4 proofs use the new rules and pricing; v3 remains accepted during the transition.
• 26,925 (Y), expected 15 September: the intelligence multiplier applies to v3 as well as v4, alongside a one-off difficulty adjustment.
• 27,615 (Z), expected 20 September: the transition ends. V3 proofs are no longer accepted.
Dates are estimates in UTC. Activation follows block height, so complete your upgrade ahead of the relevant height. Do not wait for the v3 sunset to upgrade.
Release and upgrade instructions:
https://git.tensorcash.org/tensorcash/tensorcash/releases/tag/v1.2.0
Binaries and wallets will be available on the release page once CI builds them.
The method, research and charts:
https://tensorcash.org/blog/proof-v4-the-intelligence-multiplier/
https://git.tensorcash.org/tensorcash/tensorcash/releases/tag/v1.2.0
Following yesterday's announcement, we are releasing v1.2.0: the intelligence multiplier and a substantial upgrade to consensus. Difficulty that reflects how the model generates.
Who should upgrade
• Pools, node operators and verification providers: upgrade today, 14 September, before activation at block 26,825.
• Pool miners: wait for your pool to confirm its upgrade, then follow its instructions to update your miner.
• Solo miners: upgrade as soon as possible.
Everyone must complete their upgrade before block 26,925 (Y), expected Tuesday 15 September. Pool miners should coordinate with their pool to meet this deadline.
Mainnet rollout
• 26,825 (X), expected overnight 14–15 September: Proof v4 activates. V4 proofs use the new rules and pricing; v3 remains accepted during the transition.
• 26,925 (Y), expected 15 September: the intelligence multiplier applies to v3 as well as v4, alongside a one-off difficulty adjustment.
• 27,615 (Z), expected 20 September: the transition ends. V3 proofs are no longer accepted.
Dates are estimates in UTC. Activation follows block height, so complete your upgrade ahead of the relevant height. Do not wait for the v3 sunset to upgrade.
Release and upgrade instructions:
https://git.tensorcash.org/tensorcash/tensorcash/releases/tag/v1.2.0
Binaries and wallets will be available on the release page once CI builds them.
The method, research and charts:
https://tensorcash.org/blog/proof-v4-the-intelligence-multiplier/
https://git.tensorcash.org/tensorcash/tensorcash/releases/tag/v1.2.0
TensorCash Git
TensorCash v1.2.0 - tensorcash/tensorcash
TensorCash v1.2.0 introduces proof v4 verification and miner checks for the coordinated mainnet transition.
**Binary builds are in progress. No downloadable binaries have been validated for this release yet.** Source archives are available from the tag;…
**Binary builds are in progress. No downloadable binaries have been validated for this release yet.** Source archives are available from the tag;…
👍2❤1
⚠️ TensorCash 1.2.0 activation today at block 26,925, expected between 14:00 and 18:00 UTC.
Please update your core node, wallet, explorer and web-wallet backend now. Pools and exchanges have upgraded or are upgrading now. Nodes that are not upgraded will fork at block 26,925.
Run v1.2.0a (binaries and images on the release page):
https://git.tensorcash.org/tensorcash/tensorcash/releases/tag/v1.2.0a
What happens at 26,925: the intelligence multiplier starts pricing v3 and v4 proofs, with a one-off difficulty adjustment. v3 proofs stop being accepted at block 27,615, around 20 September. Activation follows block height; the times above are estimates from the current block pace.
https://git.tensorcash.org/tensorcash/tensorcash/releases/tag/v1.2.0a
Please update your core node, wallet, explorer and web-wallet backend now. Pools and exchanges have upgraded or are upgrading now. Nodes that are not upgraded will fork at block 26,925.
Run v1.2.0a (binaries and images on the release page):
https://git.tensorcash.org/tensorcash/tensorcash/releases/tag/v1.2.0a
What happens at 26,925: the intelligence multiplier starts pricing v3 and v4 proofs, with a one-off difficulty adjustment. v3 proofs stop being accepted at block 27,615, around 20 September. Activation follows block height; the times above are estimates from the current block pace.
https://git.tensorcash.org/tensorcash/tensorcash/releases/tag/v1.2.0a
TensorCash Git
TensorCash v1.2.0a - tensorcash/tensorcash
TensorCash v1.2.0a fixes verifier and miner issues found after v1.2.0.
- Fix GPU verification of v3 proofs during the Y-to-Z pricing window by reading replay membership inputs on the host.
- Apply the native anti-parrot gate before miner submission on both…
- Fix GPU verification of v3 proofs during the Y-to-Z pricing window by reading replay membership inputs on the host.
- Apply the native anti-parrot gate before miner submission on both…
TensorCash v1.2.0b is out. We are adding 25 blocks to the transition so proof v3 hands over to v4 without a major loss of mining work.
What changes
• Pricing activation (Y) moves from block 26,925 to 26,950. X stays at 26,825 and Z stays at 27,615.
• The original v3 sampling tolerance is restored, before and after activation. The v4 rules are unchanged and no proof-validity check is relaxed.
• The Quick validation cache starts a fresh epoch, so affected blocks are revalidated on first start.
What to do
• If you updated to 1.2.0 or 1.2.0a, update to 1.2.0b.
• If you have not upgraded yet, update directly to 1.2.0b.
• Core node, wallet, explorer and web-wallet backend all apply. Miners with an explicit height override must set Y=26950.
• Nodes that are not upgraded will fork at block 26,950.
Release (binaries are being built and appear on the page as CI finishes; Windows packages are unsigned, as with 1.2.0a):
https://git.tensorcash.org/tensorcash/tensorcash/releases/tag/v1.2.0b
https://git.tensorcash.org/tensorcash/tensorcash/releases/tag/v1.2.0b
What changes
• Pricing activation (Y) moves from block 26,925 to 26,950. X stays at 26,825 and Z stays at 27,615.
• The original v3 sampling tolerance is restored, before and after activation. The v4 rules are unchanged and no proof-validity check is relaxed.
• The Quick validation cache starts a fresh epoch, so affected blocks are revalidated on first start.
What to do
• If you updated to 1.2.0 or 1.2.0a, update to 1.2.0b.
• If you have not upgraded yet, update directly to 1.2.0b.
• Core node, wallet, explorer and web-wallet backend all apply. Miners with an explicit height override must set Y=26950.
• Nodes that are not upgraded will fork at block 26,950.
Release (binaries are being built and appear on the page as CI finishes; Windows packages are unsigned, as with 1.2.0a):
https://git.tensorcash.org/tensorcash/tensorcash/releases/tag/v1.2.0b
https://git.tensorcash.org/tensorcash/tensorcash/releases/tag/v1.2.0b
TensorCash Git
TensorCash v1.2.0b - tensorcash/tensorcash
TensorCash v1.2.0b restores the original v3 sampling tolerance and starts a fresh Quick/Smell validation cache.
- Restore the v3 inverse-CDF u band to `ATOL=1e-4`, including after X and Y. The support-membership, chosen-probability binding and length checks…
- Restore the v3 inverse-CDF u band to `ATOL=1e-4`, including after X and Y. The support-membership, chosen-probability binding and length checks…
❤1