Transparency
Measured costs
What this has cost, measured from each model call's own usage block rather than estimated. The line that matters most is the one nobody enjoys publishing: work that produced nothing and was paid for anyway.
$2.5448
total measured spend
525
priced model calls
64
ledger entries
0
entries with incomplete cost
By what the money was for
| scope | spend | what it covers |
|---|---|---|
paper_read | $1.9188 | reading a paper (five stages: reader A, reader B, critic, checker, prior-art gate) |
unattributed | $0.2837 | recorded before costs carried a scope label |
reanalysis | $0.1293 | bounded re-analysis of a paper’s reported values |
checker_probe | $0.0820 | adversarial probes against the independent claim checker |
reanalysis_void | $0.0521 | re-analysis whose output was discarded by an instrument bug, and paid for anyway |
gate_variance | $0.0520 | measuring how much the prior-art verdict moves on identical evidence |
scout | $0.0269 | locating gaps and generating hypotheses |
A lower bound is still a measurement, and an unmeasured call is not. Where a call's cost could not be priced, the entry is counted and flagged rather than dropped or averaged in — the flag count is published above.
Every ledger entry (64)
| seq | scope | calls | spend | note |
|---|---|---|---|---|
| 1418 | reanalysis | 2 | $0.0192 | H51 analyst + verifier |
| 1414 | scout | 2 | $0.0044 | scout proposal + prior-art gate calls |
| 1407 | scout | 2 | $0.0085 | scout proposal + prior-art gate calls |
| 1393 | paper_read | 5 | $0.0481 | all calls measured; 5 call(s) read from cache with cached tokens priced at zero in the rate card, so the total is a lower bound |
| 1368 | paper_read | 5 | $0.0378 | all calls measured; 5 call(s) read from cache with cached tokens priced at zero in the rate card, so the total is a lower bound |
| 1343 | paper_read | 5 | $0.0439 | all calls measured; 5 call(s) read from cache with cached tokens priced at zero in the rate card, so the total is a lower bound |
| 1318 | paper_read | 5 | $0.0512 | all calls measured; 5 call(s) read from cache with cached tokens priced at zero in the rate card, so the total is a lower bound |
| 1293 | paper_read | 5 | $0.0365 | all calls measured; 5 call(s) read from cache with cached tokens priced at zero in the rate card, so the total is a lower bound |
| 1263 | paper_read | 5 | $0.0299 | all calls measured; 5 call(s) read from cache with cached tokens priced at zero in the rate card, so the total is a lower bound |
| 1238 | paper_read | 5 | $0.0365 | all calls measured; 5 call(s) read from cache with cached tokens priced at zero in the rate card, so the total is a lower bound |
| 1213 | paper_read | 5 | $0.0464 | all calls measured; 5 call(s) read from cache with cached tokens priced at zero in the rate card, so the total is a lower bound |
| 1188 | paper_read | 5 | $0.0406 | all calls measured; 5 call(s) read from cache with cached tokens priced at zero in the rate card, so the total is a lower bound |
| 1163 | paper_read | 5 | $0.0499 | all calls measured; 5 call(s) read from cache with cached tokens priced at zero in the rate card, so the total is a lower bound |
| 1138 | paper_read | 5 | $0.0570 | all calls measured; 5 call(s) read from cache with cached tokens priced at zero in the rate card, so the total is a lower bound |
| 1113 | paper_read | 5 | $0.0359 | all calls measured; 5 call(s) read from cache with cached tokens priced at zero in the rate card, so the total is a lower bound |
| 1088 | paper_read | 5 | $0.0352 | all calls measured; 5 call(s) read from cache with cached tokens priced at zero in the rate card, so the total is a lower bound |
| 1063 | paper_read | 5 | $0.0622 | all calls measured; 5 call(s) read from cache with cached tokens priced at zero in the rate card, so the total is a lower bound |
| 1038 | paper_read | 5 | $0.0291 | all calls measured; 5 call(s) read from cache with cached tokens priced at zero in the rate card, so the total is a lower bound |
| 1013 | paper_read | 5 | $0.0493 | all calls measured; 5 call(s) read from cache with cached tokens priced at zero in the rate card, so the total is a lower bound |
| 988 | paper_read | 5 | $0.0448 | all calls measured; 5 call(s) read from cache with cached tokens priced at zero in the rate card, so the total is a lower bound |
| 963 | paper_read | 5 | $0.0339 | all calls measured; 5 call(s) read from cache with cached tokens priced at zero in the rate card, so the total is a lower bound |
| 938 | paper_read | 5 | $0.0467 | all calls measured; 5 call(s) read from cache with cached tokens priced at zero in the rate card, so the total is a lower bound |
| 913 | paper_read | 5 | $0.0363 | all calls measured; 5 call(s) read from cache with cached tokens priced at zero in the rate card, so the total is a lower bound |
| 888 | paper_read | 5 | $0.0365 | all calls measured; 5 call(s) read from cache with cached tokens priced at zero in the rate card, so the total is a lower bound |
| 863 | paper_read | 5 | $0.0630 | all calls measured; 5 call(s) read from cache with cached tokens priced at zero in the rate card, so the total is a lower bound |
| 838 | paper_read | 5 | $0.0449 | all calls measured; 5 call(s) read from cache with cached tokens priced at zero in the rate card, so the total is a lower bound |
| 813 | paper_read | 5 | $0.0378 | all calls measured; 5 call(s) read from cache with cached tokens priced at zero in the rate card, so the total is a lower bound |
| 788 | paper_read | 5 | $0.0316 | all calls measured; 5 call(s) read from cache with cached tokens priced at zero in the rate card, so the total is a lower bound |
| 763 | paper_read | 5 | $0.0421 | all calls measured; 5 call(s) read from cache with cached tokens priced at zero in the rate card, so the total is a lower bound |
| 738 | paper_read | 5 | $0.0320 | all calls measured; 5 call(s) read from cache with cached tokens priced at zero in the rate card, so the total is a lower bound |
| 713 | paper_read | 5 | $0.0424 | all calls measured; 5 call(s) read from cache with cached tokens priced at zero in the rate card, so the total is a lower bound |
| 688 | paper_read | 5 | $0.0401 | all calls measured; 5 call(s) read from cache with cached tokens priced at zero in the rate card, so the total is a lower bound |
| 663 | paper_read | 5 | $0.0478 | all calls measured; 5 call(s) read from cache with cached tokens priced at zero in the rate card, so the total is a lower bound |
| 638 | paper_read | 5 | $0.0520 | all calls measured; 5 call(s) read from cache with cached tokens priced at zero in the rate card, so the total is a lower bound |
| 613 | paper_read | 5 | $0.0622 | all calls measured; 5 call(s) read from cache with cached tokens priced at zero in the rate card, so the total is a lower bound |
| 588 | paper_read | 5 | $0.0285 | all calls measured; 5 call(s) read from cache with cached tokens priced at zero in the rate card, so the total is a lower bound |
| 563 | paper_read | 5 | $0.0372 | all calls measured; 5 call(s) read from cache with cached tokens priced at zero in the rate card, so the total is a lower bound |
| 538 | paper_read | 5 | $0.0307 | all calls measured; 5 call(s) read from cache with cached tokens priced at zero in the rate card, so the total is a lower bound |
| 513 | paper_read | 5 | $0.0740 | all calls measured; 5 call(s) read from cache with cached tokens priced at zero in the rate card, so the total is a lower bound |
| 488 | paper_read | 5 | $0.0297 | all calls measured; 5 call(s) read from cache with cached tokens priced at zero in the rate card, so the total is a lower bound |
| 463 | paper_read | 5 | $0.0363 | all calls measured; 5 call(s) read from cache with cached tokens priced at zero in the rate card, so the total is a lower bound |
| 408 | paper_read | 5 | $0.0468 | all calls measured; 5 call(s) read from cache with cached tokens priced at zero in the rate card, so the total is a lower bound |
| 376 | paper_read | 5 | $0.0490 | all calls measured; 5 call(s) read from cache with cached tokens priced at zero in the rate card, so the total is a lower bound |
| 351 | paper_read | 5 | $0.0690 | all calls measured; 5 call(s) read from cache with cached tokens priced at zero in the rate card, so the total is a lower bound |
| 326 | paper_read | 5 | $0.0517 | all calls measured; 5 call(s) read from cache with cached tokens priced at zero in the rate card, so the total is a lower bound |
| 301 | paper_read | 5 | $0.0443 | all calls measured; 5 call(s) read from cache with cached tokens priced at zero in the rate card, so the total is a lower bound |
| 276 | paper_read | 5 | $0.0383 | all calls measured; 3 call(s) read from cache with cached tokens priced at zero in the rate card, so the total is a lower bound |
| 245 | cost_correction → corrects reanalysis | 0 | $-0.0013 | recorded $0.111391; measured $0.110121 (main run $0.100981 + H24/H30 rerun $0.009140) |
| 244 | cost_correction → corrects reanalysis_void | 0 | $-0.0759 | recorded $0.128007 by summing overlapping per-record usages; the run measured $0.052062 |
| 243 | cost_correction → corrects cost_correction | 0 | $0.0772 | reverse the combined correction; the same money is re-applied below against the scopes that were wrong |
| 242 | cost_correction | 0 | $-0.0772 | negative correction: the re-analysis per-record usage figures double-counted because usage was attributed with sumUsage(usages.slice(-2)) over a run-wide accumu |
| 241 | reanalysis_void | 15 | $0.1280 | backfilled by cost reconciliation: summed from the 7 void re-analysis records themselves; the output was discarded by a property-name bug but the calls were pai |
| 240 | reanalysis | 4 | $0.0104 | backfilled by cost reconciliation: run-reanalysis.mjs --only H24,H30 re-run to correct the recorded reason |
| 239 | reanalysis | 14 | $0.1010 | backfilled by cost reconciliation: run-reanalysis.mjs first successful run: 7 hypotheses x 2 passes, $0.100981 |
| 238 | gate_variance | 13 | $0.0520 | backfilled by cost reconciliation: scripts/gate-variance.mjs run, 2026-09-14: 13 adjudicator calls, $0.052 (commit ba8e352) |
| 237 | checker_probe | 215 | $0.0820 | backfilled by cost reconciliation: scripts/checker-probe.mjs run, 2026-09-14: 215 probes, $0.082 (commit e9e6337) |
| 236 | scout | 3 | $0.0140 | backfilled by cost reconciliation: run-scout.mjs final cost lines, both scout runs plus the gate: $0.0067 + $0.0073 measured by the runs themselves |
| 208 | — | 5 | $0.0398 | all calls measured; 5 call(s) read from cache with cached tokens priced at zero in the rate card, so the total is a lower bound |
| 183 | — | 5 | $0.0407 | all calls measured; 5 call(s) read from cache with cached tokens priced at zero in the rate card, so the total is a lower bound |
| 134 | — | 5 | $0.0470 | all calls measured; 5 call(s) read from cache with cached tokens priced at zero in the rate card, so the total is a lower bound |
| 109 | — | 5 | $0.0438 | all calls measured; 5 call(s) read from cache with cached tokens priced at zero in the rate card, so the total is a lower bound |
| 70 | — | 5 | $0.0353 | all calls measured; 5 call(s) read from cache with cached tokens priced at zero in the rate card, so the total is a lower bound |
| 46 | — | 5 | $0.0360 | all calls measured; 5 call(s) read from cache with cached tokens priced at zero in the rate card, so the total is a lower bound |
| 22 | — | 5 | $0.0411 | all calls measured; 5 call(s) read from cache with cached tokens priced at zero in the rate card, so the total is a lower bound |
Machine-readable: /society/costs.json.