TensorCash
Carteira TSCParticipe

Escolha um idioma

For miners and verifiers

What you earn

The block reward

A miner that finds a block receives the block reward and the transaction fees in that block [1]. The reward follows a fixed schedule [2].

  • The first epoch lasts 715 blocks, about five days at ten minutes per block. Each next epoch is twice as long, up to 732,160 blocks.
  • The reward starts at 715 $TSC per block and is multiplied by three fifths every epoch, rounded down to the smallest unit.
  • After 48 epochs the total supply ends at 21,184,153.03530240 $TSC.

Since block 22,165 the reward is 55.5984 $TSC per block. From block 45,045 it will be 33.35904 $TSC [2][3] (see Emission).

A new block reward stays locked until 100 more blocks have been added to the chain, as in Bitcoin. After that, the miner can use it. Pools pay out once that point is reached [4].

Payment for inference

Customers pay for answers through apps such as AskTilly and Bob Labs. The app pays the pool or provider that served the request, which passes the payment on to its miners. The Bob Labs chat, for example, charges 1 $TSC per million output tokens, and a quarter of that for input tokens, and is served by the miners in its pool [4][5].

How the difficulty adjusts

The network aims for one block every ten minutes [6]. Since block 20,440 the difficulty adjusts with every block, using a rule called ASERT [9]. When blocks arrive faster than this schedule, the difficulty rises a little, and when they arrive slower, it goes down. Until block 46,738, it responds with a half-life of 12 hours, and after that more gradually, with a half-life of 48 hours [9]. Before block 20,440, it was recalculated in steps every 2,016 blocks, the same interval Bitcoin uses.

Each block header holds two targets. One is the general target of the network. The other is adjusted for the model that produced the proof, and a block only counts if its hash falls below that adjusted target. Larger models get an easier target, and smaller, cheaper models a harder one, so switching to a lighter model does not make blocks easier to find [2][7].

The verifier also gives each proof an intelligence multiplier between 1× and 12×, depending on the entropy of the 256 steps in the window. The target is divided by the multiplier, so at 2× a block takes about twice as many attempts to find. Miners cannot choose the multiplier, and the block reward stays the same [8].

Protecting future income

A miner's income falls when the difficulty rises, because the same GPUs then find fewer blocks. Compute derivatives let a miner take a position that pays out when the difficulty rises, which makes up for that loss (see Compute derivatives).

References

  1. [1]TensorCash whitepaper.tensorcash.org/whitepapers/TensorCash_Whitepaper-modern.pdf
  2. [2]Core-Node whitepaper.tensorcash.org/whitepapers/CoreNode_Whitepaper-modern.pdf
  3. [3]TensorCash blog, Welcome to the Bitcoin genesis era, 13 August 2026.tensorcash.org/blog/welcome-to-the-bitcoin-genesis-era/
  4. [4]TensorCash website, Mining pools.tensorcash.org/pools
  5. [5]TensorCash blog, The reply that mines the chain, 9 August 2026.tensorcash.org/blog/third-party-ecosystem/
  6. [6]TensorCash blog, COMPUTE: options on the network difficulty, 19 July 2026.tensorcash.org/blog/tokenised-difficulty-options/
  7. [7]TensorCash blog, One chain, many models: how TensorCash runs multiple AI models, 14 August 2026.tensorcash.org/blog/how-tensorcash-runs-multiple-models/
  8. [8]TensorCash blog, Proof v4: the intelligence multiplier, 13 September 2026.tensorcash.org/blog/proof-v4-the-intelligence-multiplier/
  9. [9]TensorCash v1.1.0 release notes, 3 August 2026.git.tensorcash.org/tensorcash/tensorcash/releases/tag/v1.1.0