ModelCensusopen-source ai reliability harness
The finding · Class 7 · Temporal & Knowledge Boundary

Real context moved 0.0pp on Date-Arithmetic Error.

Better than the control, worse than nothing
The figure above is measured against the irrelevant-document control. Against no context at all this card is worse: 23% bare against 25% with the best supplied source. What the source repaired is damage the presence of a document did.

~intervals overlap — not separated from no change

Given an injected "current date," the model computes relative dates/intervals incorrectly. The rule

qwen/qwen-2.5-72b-instructfmi_7_35 repeatsfmi_7_3_v1
The modeAll 7 cards
In one frame
screenshot this
Date-Arithmetic Errorfmi_7_3
qwen/qwen-2.5-72b-instruct
Same question, both times
Today is 2026-03-10. What date is 30 days from today? Answer with the full date.
No contextPASS

The date 30 days from today, which is 2026-03-10, is 2026-04-09.

detector fmi_7_3_v1: date arithmetic correct

General contextPASS

The date 30 days from today, 2026-03-10, is 2026-04-09.

detector fmi_7_3_v1: date arithmetic correct

Verdict unchanged on this case — the aggregate below is what carries the effect. Across 5 repeats: 0.0pp (intervals overlap).fmi_7_3_v1 · qwen/qwen-2.5-72b-instruct · 2026-08-15 · ground truth as of 2026-08-09

loop 2026-08-15 · qwen/qwen-2.5-72b-instruct as served · open weights · released 2024-09 · 72B · set s1.2-live · cases v1.4 · conditions v1.0 · ground truth as of 2026-08-09

From no context to real context
5 repeats per case
failure rate · left is fewer failures0%15%30%45%60%No context23%[12%–38%] n=40Irrelevant context25%presence — a document being there[14%–40%] n=40+2.5pp OVERLAPSGeneral context25%content — the information itself[14%–40%] n=400.0pp OVERLAPS
whiskers are 95% intervals · each step is measured from the stop before it, not from zero
Try the other contexts
stored · no model call
1Questionheld fixed
Today is 2026-03-10. What date is 30 days from today? Answer with the full date.

expected as of 2026-08-09

From 2026-03-10, 30 days later is 2026-04-09 (9 April 2026). Thirty days from an ordinary date.

2Contextchange me

Nothing supplied. The baseline.

3ResponseNo context
PASS

The date 30 days from today, which is 2026-03-10, is 2026-04-09.

fmi_7_3_v1

detector fmi_7_3_v1: date arithmetic correct

The scoring
Failure rate by context
No context23% [12%–38%] n=40
Irrelevant context25% [14%–40%] n=40
General context25% [14%–40%] n=40
Curated context25% [14%–40%] n=40

0–60% · whisker = 95% interval

Trap vs answerable
No contexttrap 36%answerable 0%
Irrelevant contexttrap 40%answerable 0%
General contexttrap 40%answerable 0%
Curated contexttrap 40%answerable 0%

never pooled · a trap ratio would skew the delta

Residual — what survives

Unchanged by context: 23% with none, 25% with the strongest supplied source. For a control mode that flat line is the result — it is what licenses reading movement elsewhere in this loop as grounding rather than as a document being present.

Mitigations from the index — the claim
  • date libraries via tools
  • explicit date arithmetic

fmi_7_3_v1 · 5 repeats per case · best context general context · rollups + residuals committed · probe log outside git · manifest hash ties them · method