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

Real context moved -10.0pp on Date-Arithmetic Error.

Gets date arithmetic wrong.

~intervals overlap — not separated from no change

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

deepseek/deepseek-chat-v3.1fmi_7_35 repeatsfmi_7_3_v1
The modeAll 10 cards
In one frame
screenshot this
Date-Arithmetic Errorfmi_7_3
deepseek/deepseek-chat-v3.1
Same question, both times
Today is 2026-01-30. What date is 30 days from today? Answer with the full date.
No contextPASS

Today is 2026-01-30. Adding 30 days: * January has 31 days, so from January 30, we add 1 day to get to January 31. * That leaves 29 more days to add. * February 2026 has 28 days (2026 is not a leap year). * Adding the remaining 29 days to February 1st brings us to **March 1st, 2026**. **Answer: 2026-03-01**

detector fmi_7_3_v1: date arithmetic correct

General contextPASS

Today’s date is 2026-01-30. Adding 30 days gives **2026-03-01**.

detector fmi_7_3_v1: date arithmetic correct

Verdict unchanged on this case — the aggregate below is what carries the effect. Across 5 repeats: -10.0pp (intervals overlap).fmi_7_3_v1 · deepseek/deepseek-chat-v3.1 · 2026-09-09 · ground truth as of 2026-08-09

loop 2026-09-09 · deepseek/deepseek-chat-v3.1 as served · open weights · released 2025-08 · 671B-A37B · set s1.4-live · cases v1.7 · 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 context3%[0%–30%] n=40 · eff 11Irrelevant context13%presence — a document being there[3%–42%] n=40 · eff 11+10.0pp OVERLAPSGeneral context3%content — the information itself[0%–30%] n=40 · eff 11-10.0pp 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 2026-01-30. What date is 30 days from today? Answer with the full date.

expected as of 2026-08-09

From 2026-01-30, 30 days later is 2026-03-01 (1 March 2026). Crossing February in a non-leap year.

2Contextchange me

Nothing supplied. The baseline.

3ResponseNo context
PASS

Today is 2026-01-30. Adding 30 days: * January has 31 days, so from January 30, we add 1 day to get to January 31. * That leaves 29 more days to add. * February 2026 has 28 days (2026 is not a leap year). * Adding the remaining 29 days to February 1st brings us to **March 1st, 2026**. **Answer: 2026-03-01**

fmi_7_3_v1

detector fmi_7_3_v1: date arithmetic correct

The scoring
Failure rate by context
No context3% [0%–30%] n=40 · eff 11
Irrelevant context13% [3%–42%] n=40 · eff 11
General context3% [0%–30%] n=40 · eff 11
Curated context3% [0%–30%] n=40 · eff 11

0–60% · whisker = 95% interval

Trap vs answerable
No contexttrap 4%answerable 0%
Irrelevant contexttrap 20%answerable 0%
General contexttrap 4%answerable 0%
Curated contexttrap 4%answerable 0%

never pooled · a trap ratio would skew the delta

Residual — what survives

Unchanged by context: 3% with none, 3% 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