Skip to content
dfpn. Read the protocol

Tokenomics

The economics of an honest verdict.

DFPN is a Solana SPL token. It is used for staking, rewards, protocol fees (with an optional discount versus paying in SOL), and governance weight. Every number on this page is taken directly from the dfpn README and the tokenomics document in the repo; we have not invented multipliers, APYs, or yields.

Total supply

1,000,000,000

DFPN, fixed

Token type

SPL

on Solana

Network reward pool

38%

of total supply

Fee currency

SOL

DFPN discount optional

Supply allocation

The reward pool is the largest single allocation, sized to bootstrap worker and model supply through the first several years. Treasury, ecosystem, and team allocations cover ops, integrations, and contributor compensation.

  • Network rewards (workers + model developers) 38%
  • Treasury (audits, grants, ops, insurance) 20%
  • Team & advisors (4-year vesting, 1-year cliff) 18%
  • Ecosystem growth (partnerships, integrations, data grants) 12%
  • Strategic backers (2-year lock, linear vesting) 7%
  • Liquidity & market making 5%

Fee distribution per request

Every paid request is split four ways. The split is governed and can be adjusted by token-holder governance, but the baseline is documented and stable.

  • Workers 65%
  • Model developers 20%
  • Treasury 10%
  • Insurance pool 5%

Fee tiers can vary by modality (video > image > audio) and SLA.

Emissions schedule

Emissions are carved out of the 38% reward allocation and decline over eight years. Governance may enable up to a 1% annual inflation tail if request fees alone do not sustain the worker market.

  • Year 1 12% of total supply
  • Year 2 9% of total supply
  • Year 3 6% of total supply
  • Year 4 4% of total supply
  • Year 5 3% of total supply
  • Year 6 2% of total supply
  • Year 7 1% of total supply
  • Year 8 1% of total supply

Optional tail: up to 1% annual inflation under governance.

Stake requirements

Stakes scale with demand: the higher the median request fee, the higher the floor. The static starting values below are tunable by governance.

Role Floor Scaling rule
Workers 5,000 DFPN >= 30x median request fee per epoch
Model developers 20,000 DFPN per model version >= 150x median request fee per epoch
Challengers 5% of disputed reward Posted at the time of challenge

Minimum request fee starts at 0.02 SOL (or the DFPN equivalent), again tunable by governance.

Slashing schedule

A short challenge window precedes any slashing event. Governance can override in exceptional cases. The design intent is to make repeated dishonesty far more expensive than a single honest mistake.

Invalid results

10% stake slash

Fraud or collusion

25% to 50% slash + temporary ban

Missed deadlines

1% to 3% slash + reputation decay

Repeated low-quality results

Progressive slashing and retirement

Reward scoring weights

Per-epoch rewards are weighted by measurable performance. Stake weight enters the calculation but is capped, so participants cannot simply buy their way to the top of the leaderboard.

  • Accuracy · Agreement with benchmark or consensus 50%
  • Availability · Uptime and response responsiveness 25%
  • Latency · Timeliness vs request deadline 15%
  • Consistency · Stability of performance across epochs 10%

Governance controls

Token holders can adjust fee splits, stake minimums, emission cadence, and slashing ranges. They also approve new benchmark datasets and trigger rotation of hidden test sets — a quietly important power, since static benchmarks degrade as detectors are trained against them.

Protocol parameters

Fee splits, stake minimums, emission schedule, slashing ranges.

Benchmarks

Approve and rotate datasets used for periodic evaluation of models.

Inflation tail

Enable or disable the optional 1% annual tail if fee revenue is thin.

Numbers, not vibes.

The figures above are the ones the dfpn repo commits to today. Anything not listed — market price, circulating supply at TGE, partner integrations — is intentionally absent: when those numbers exist they will live in the repo first, here second.