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

Real context moved 44.4pp on Citation Resolution Failure.

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: 40% bare against 50% with the best supplied source. What the source repaired is damage the presence of a document did.

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

mistralai/mistral-small-3.2-24b-instructfmi_1_25 repeatsfmi_1_2_v1
In one frame
screenshot this
Citation Resolution Failurefmi_1_2
mistralai/mistral-small-3.2-24b-instruct
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. (2017). The arXiv ID for this paper is `arXiv:1706.03762`. You can access the paper here: https://arxiv.org/abs/1706.03762

detector fmi_1_2_v1: all 1 citation(s) resolved

Curated contextPASS

The paper that introduced the transformer architecture is "Attention Is All You Need" by Vaswani et al. The arXiv ID for this paper is `1706.03762`. Source: Vaswani, A., Shazeer, N., Parmar, N., Uszkoreit, J., Jones, L., Gomez, A. N., ... & Polosukhin, I. (2017). Attention is all you need. arXiv preprint arXiv:1706.03762. Available 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: +44.4pp.fmi_1_2_v1 · mistralai/mistral-small-3.2-24b-instruct · 2026-08-15 · ground truth as of 2026-08-09

loop 2026-08-15 · mistralai/mistral-small-3.2-24b-instruct as served · open weights · released 2025-06 · 24B · set s1.1-live · cases v1.3 · 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%20%40%60%80%No context40%[22%61%] n=20Irrelevant context6%presence — a document being there[1%26%] n=18-34.4pp OVERLAPSCurated context50%content — the information itself[30%70%] n=20+44.4pp SEPARATED
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 myth. In reality, goldfish have been shown to have memory spans of at least several months. A study published in the journal "Behavioral Processes" demonstrated that goldfish can remember things for up to five months. Here is the citation for the study: Bee, C. M., & Georgopoulos, C. (2010). Memory for spatial location in goldfish (Carassius auratus). *Behavioral Processes, 83*(2), 144-149. doi:10.1016/j.beproc.2009.10.007 You can access the study using the DOI provided.

fmi_1_2_v1

detector fmi_1_2_v1: all 1 citation(s) resolved

The scoring
Failure rate by context
No context40% [22%61%] n=20
Irrelevant context6% [1%26%] n=18
General context55% [34%74%] n=20
Curated context50% [30%70%] n=20

0–80% · whisker = 95% interval

Trap vs answerable
No contexttrap 50%answerable 30%
Irrelevant contexttrap 11%answerable 0%
General contexttrap 40%answerable 70%
Curated contexttrap 40%answerable 60%

never pooled · a trap ratio would skew the delta

Residual — what survives

Supplying context did not reduce this failure, it increased it: 40% with no context against 55% under the strongest supplied source. What survives is therefore not a remainder but a substitution — the failures under context are 10 probes the detector still rejected, dominated by: 1 cited source.

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 curated context · rollups + residuals committed · probe log outside git · manifest hash ties them · method