Fulfillment ledger
Each question carries its current state, the sections that answer it, and what is still open. A question that is answered in research custody but not carried onto this site says so.
Answered
The posted range across comparable models is 111x, and the quality-adjusted spread is wider still — but effective cost is set by output volume, retries, and cache economics, not by the rate card.
Still open
An effective-cost calculator. Board-gated tooling, not blocked on research.
Answered, and modeled in the repository
It depends on utilization and procurement scope, and the crossover band is quantified per lane rather than argued.
Still open
A public-facing version of the crossover model, and a quarterly price refresh.
Answered
Route by measured effective cost per successful task at your quality bar; the savings are real and quality-conditioned, and the routing fee is thin.
Still open
A routing decision tool.
Answered
The compute is cheap and everything around it is not; the substitute-or-complement question with retrieval decides it before the cost model does.
Still open
A payback calculator.
Answered structurally, anchored on the cost side
Answered structurally with both anchors: cost-side — frontier list pricing runs below sustainable infrastructure return (verified lab margins), so posted prices are effectively investor-subsidized; channel-side — zero-price inference is structurally embedded at scale (28+ :free variants; the largest usage study excludes free-launch access from rankings). The aggregate user-side fraction remains unmeasured.
Still open
An aggregate fraction, if one is ever published. Zero-price channel caps rechecked quarterly.
Answered as six dated bets
Not as a curve. Six binary bets, each with a probability, an evidence grade, and a condition that annuls it.
Still open
Resolution scoring as the dates arrive.
Answered in research custody; carried onto this site only as a forecast bet
Quantization studies and current-generation NPU throughput baselines are held in the research notebooks. The public report carries the edge question as a dated bet rather than a decision surface.
Still open
An edge feasibility route on this site, and the tool behind it.
Answered for the coupled pairs evidence supports
Four couplings are now evidenced, not just named: (1) spot-risk pricing against on-prem ownership (a 20% interruption premium moves the ownership crossover from ~67% to ~47% utilization); (2) subsidy contraction against the control premium — if FC4 resolves YES, owned/sovereign infrastructure appreciates as a hedge; (3) model efficiency against hardware supply — the price-decline attribution split (FC1) determines whether Rubin-class hardware moves the TCO bands at all; (4) procurement scope against every ownership crossover — GPU-only vs node-loaded capex changes answers from ~21% to ~57%+ utilization. The full n-way interaction remains open research.
Still open
A cross-dimensional interaction study. Open research, not a publication gap.
Answered
Six sequential decisions: assume unit cost falls while total spend rises; treat subsidies as expiring and budget at list price; decide capacity ownership by procurement scope; keep the routing stack portable; budget evaluation explicitly; write reopen conditions into the budget itself.
Still open
Aggregate subsidized-fraction data (FC4 boundary); FC resolutions begin H2 2026+.