Bittensor’s runtime 467 extends the fee changes introduced in runtime 466. The new code halves the coefficients used to price native transaction weight and encoded length, lowers the EVM base-fee setting, and introduces a smaller billing allowance for Root claims.
These changes address different parts of a transaction bill. For subnet operators and TAO holders, the useful question is which component their operation pays for.
Two native coefficients fall
The weight coefficient falls from 0.0005 to 0.00025 rao per unit of reference time. Reference time is the runtime’s accounting measure for computational work. The encoded-length coefficient falls from one rao per byte to half a rao per byte.
Together, those parameters lower the price attached to the same billable weight and transaction length. The final amount still depends on the call, its fee treatment and other components such as tips.
Runtime 467 also lowers the EVM default base-fee parameter from 20 to 10 gwei. The accompanying test changes the quiet-block expectation from 10 to 5 gwei. That is a separate mechanism from native transaction pricing.
Root claims get a billing allowance
A Root claim can inspect hotkeys and basket holdings. The code now calculates its subsidized fee weight using four claim units and an associated allowance for queued-deposit work.
The execution envelopes remain 256 units for a coldkey-wide claim and 129 for a single-hotkey claim, plus the flush allowance. A cheaper billing calculation therefore leaves the broader work budget intact.
For holders claiming small amounts, the relationship between accrued rewards and claim costs matters. This revision changes the cost side of that calculation. Each holder’s result depends on the actual claim and fees.
Selected calls have an additional cap
The runtime includes a table of fee-weight caps based on declared weights from runtime 459. Listed operations include several stake movements, registrations, key changes and Root claims.
The calculation first applies the relevant scan and claim discounts, then takes the lower value against the listed cap. The cap targets reference-time pricing. Its scope is the explicit call list.
Supported batch and proxy wrappers receive discounts for eligible inner calls. The code preserves separate admission and execution accounting, keeping the fee subsidy from freeing additional block capacity.
Runtime 466’s staking scan adjustment was narrower. Runtime 467 adds several billing changes around that foundation, making the composition of the fee more important than any single percentage headline.
Verification
Tao Outsider read the official v466-to-v467 implementation diff, including the fee coefficients, Root allowance, selected-call caps and payment-wrapper changes. Associated tests were reviewed as source evidence; they were not executed in this reporting pass. Actual user savings require matched transaction estimates or measurements. This report makes no AMM trading-fee, throughput or completed alpha-refund claim.
The Blog observed specVersion 467 at a finalized Finney block on September 18 at 19:17 Lisbon time. GitHub still labelled v467 proposed and prerelease during the September 19 check. The saved RPC observation establishes a version at a specific block; precise activation time and deployed bytecode equivalence remain outside this verification.
Sources
Official v466 to v467 comparison
Runtime configuration and EVM fee parameter
Root allowances and fee-weight caps
Was this article useful?
One tap feedback helps us improve each post.
