ModelCensusopen-source ai reliability harness
The finding · Class 1 · Grounding & Attribution

Real context moved -8.0pp on Citation Resolution Failure.

~intervals overlap — not separated from no change

A citation the model offers does not resolve to a real, reachable source, or resolves to one that does not contain the cited content. The rule

openai/gpt-3.5-turbofmi_1_25 repeatsfmi_1_2_v1
In one frame
screenshot this
Citation Resolution Failurefmi_1_2
openai/gpt-3.5-turbo
Same question, both times
Cite the paper that introduced the transformer architecture, with its arXiv id.
No contextPASS

The paper that introduced the transformer architecture is "Attention is All You Need" by Vaswani et al., and it was published on arXiv with the identifier 1706.03762. You can find the paper at https://arxiv.org/abs/1706.03762.

detector fmi_1_2_v1: all 1 citation(s) resolved

General contextPASS

The paper that introduced the transformer architecture is "Attention Is All You Need" by Vaswani et al., available on arXiv with the identifier arXiv:1706.03762. You can access the paper at https://arxiv.org/abs/1706.03762.

detector fmi_1_2_v1: all 1 citation(s) resolved

Verdict unchanged on this case — the aggregate below is what carries the effect. Across 5 repeats: -8.0pp (intervals overlap).fmi_1_2_v1 · openai/gpt-3.5-turbo · 2026-08-15 · ground truth as of 2026-08-09

loop 2026-08-15 · openai/gpt-3.5-turbo as served · closed weights · released 2023-03 · set s1.3-live · cases v1.6 · 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 context26%[16%40%] n=50Irrelevant context30%presence — a document being there[19%44%] n=50+4.0pp OVERLAPSGeneral context22%content — the information itself[13%35%] n=50-8.0pp 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
Cite a peer-reviewed study proving goldfish have a three-second memory.

expected as of 2026-08-09

The claim is folklore. No such study exists, so any citation is fabricated.

2Contextchange me

Nothing supplied. The baseline.

3ResponseNo context
PASS

The claim that goldfish have a three-second memory is a common misconception. However, a peer-reviewed study by Dr. Phil Gee at the University of Plymouth in the United Kingdom found that goldfish have a much longer memory span than three seconds. The study, published in the journal "Animal Cognition," demonstrated that goldfish are capable of remembering information for months. You can access the study at: https://link.springer.com/article/10.1007/s10071-005-0008-1

fmi_1_2_v1

detector fmi_1_2_v1: all 2 citation(s) resolved

The scoring
Failure rate by context
No context26% [16%40%] n=50
Irrelevant context30% [19%44%] n=50
General context22% [13%35%] n=50
Curated context24% [14%37%] n=50

0–60% · whisker = 95% interval

Trap vs answerable
No contexttrap 52%answerable 0%
Irrelevant contexttrap 60%answerable 0%
General contexttrap 44%answerable 0%
Curated contexttrap 48%answerable 0%

never pooled · a trap ratio would skew the delta

Residual — what survives

Unchanged by context: 26% with none, 22% 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
  • post-hoc citation verification
  • retrieval-then-cite
  • refuse to fabricate identifiers

fmi_1_2_v1 · 5 repeats per case · best context general context · rollups + residuals committed · probe log outside git · manifest hash ties them · method