Model version #4

Model version #4 · published 2026-08-11

First published version — no previous version to compare against.

NIST P-256 · forecast interval, P10 to P90 (calendar dates) · model version #4
P10, P50 and P90 dates for the NIST P-256 forecast
P10 (10% of futures cross by) P50 (median estimate) P90 (90% of futures cross by)
2059-10-29 Median unresolved Not defined

The bar ends at the modelled horizon, not at a P90. Percentiles shown as “not defined” fall in the censored tail: too many simulated scenarios never reach a crossing for those percentiles to be estimated, so no upper bound is drawn rather than one being placed at the horizon.

Cumulative probability of a Q-Day crossing by year-end, NIST P-256 · probability in per cent · calendar years · model version #4
0% 25% 50% 75% 100% 2026 2042 2058 2074 2086

P10 (10% of futures cross by): 10% by 2059-10-29

Each dashed rule marks the date a percentile falls on, and the dot on it shows how far the plotted curve has risen by that date. The two differ on purpose: percentiles are computed at full sample resolution, while the curve only steps at year-ends, so the curve reads the previous year-end's level on that date. Read each percentile's own level from the label or this legend, not from where its rule meets the curve.

72.31% of simulated scenarios do not reach a crossing at all within the modelled horizon, which ends in 2086. The curve therefore stops below 100% by design; the remainder is not an omission and is not extrapolated.

Cumulative crossing probability by year-end, NIST P-256 — the same values plotted above.

2026
0.00
2027
0.00
2028
0.00
2029
0.00
2030
0.00
2031
0.00
2032
0.00
2033
0.00
2034
0.00
2035
0.00
2036
0.00
2037
0.01
2038
0.01
2039
0.01
2040
0.01
2041
0.02
2042
0.02
2043
0.02
2044
0.03
2045
0.03
2046
0.03
2047
0.04
2048
0.04
2049
0.05
2050
0.05
2051
0.06
2052
0.06
2053
0.07
2054
0.07
2055
0.08
2056
0.08
2057
0.09
2058
0.09
2059
0.10

P10

2060
0.11
2061
0.11
2062
0.12
2063
0.13
2064
0.13
2065
0.14
2066
0.15
2067
0.15
2068
0.16
2069
0.17
2070
0.17
2071
0.18
2072
0.19
2073
0.20
2074
0.20
2075
0.21
2076
0.22
2077
0.22
2078
0.23
2079
0.24
2080
0.24
2081
0.25
2082
0.26
2083
0.26
2084
0.27
2085
0.28
2086
0.28

What moved

This is the first published version, so there is nothing to compare it against.

Percentile dates for this version and the previous published one.
Percentile This version Previous published Change
P10 2059-10-29
P50 Not reached within horizon
P90 Not reached within horizon

Evidence in this version

This version was computed from a frozen snapshot of 2 reviewed evidence records.

Entered the model (2)

Configuration and reproducibility

The exact inputs this version can be reproduced from.
Model versionv1
Simulation configuration v1
Evidence snapshot v1 (2 records)
Scenario count2000000
Random seed20260809
Code commit c4354dd325a3c5e3bc3d4cf7807fd4b6664c8a1e
Container image sha256:1c918f10a7aae12deb642b9888fb9b8ec353ba5572e410011305af4e264e1965

Validation

✓ Passed Validated 2026-08-10. 19 checks ran.

Approved and published by the same person. This project has one operator, so the two independent judgements a separation of duties asks for are, here, one.

Publication digest: 42fd59acd21209032efe68cb3a18709c369edbc0a09cf8d48dc1784c990aaa49 — a SHA-256 digest of the validated inputs, not a cryptographic signature by a key holder.

Validation checks reproducibility and internal consistency. It does not check whether the forecast is correct — see the methodology.

Publication history

Publication and rollback events for this version.
WhenActionNote
2026-08-11 13:25 UTC Published Approved and published by the same person. This project has one operator, so the two judgements separation of duties asks for are one judgement.

This is a probabilistic model of CRQC capability based on public evidence. Its primary result is a forecast interval built from the percentiles the evidence can resolve. A percentile that does not cross within the model’s horizon is reported as undefined, never estimated, together with the share of scenarios that do not cross. The model is recalculated weekly and does not claim day-level or second-level certainty.

Back to model history