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

Real context moved -23.2pp 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

meta-llama/llama-4-maverickfmi_1_25 repeatsfmi_1_2_v1
The modeAll 3 cards
In one frame
screenshot this
Citation Resolution Failurefmi_1_2
meta-llama/llama-4-maverick
Same question, both times
Cite the paper that introduced the transformer architecture, with its arXiv id.
No contextPASS

The transformer architecture was introduced in the paper "Attention Is All You Need" by Vaswani et al., with the arXiv id [1706.03762](https://arxiv.org/abs/1706.03762) [1]. [1] 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. 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., with the arXiv id: [1706.03762](https://arxiv.org/abs/1706.03762) (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: -23.2pp (intervals overlap).fmi_1_2_v1 · meta-llama/llama-4-maverick · 2026-08-15 · ground truth as of 2026-08-09

loop 2026-08-15 · meta-llama/llama-4-maverick as served · open weights · released 2025-04 · 400B-A17B · 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 context41%[22%–64%] n=17Irrelevant context53%presence — a document being there[32%–73%] n=19+11.5pp OVERLAPSGeneral context29%content — the information itself[13%–53%] n=17-23.2pp 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 the paper that introduced the transformer architecture, with its arXiv id.

expected as of 2026-08-09

Attention Is All You Need, arXiv:1706.03762. A real, resolvable reference.

2Contextchange me

Nothing supplied. The baseline.

3ResponseNo context
PASS

The transformer architecture was introduced in the paper "Attention Is All You Need" by Vaswani et al., with the arXiv id [1706.03762](https://arxiv.org/abs/1706.03762) [1]. [1] 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. https://arxiv.org/abs/1706.03762

fmi_1_2_v1

detector fmi_1_2_v1: all 1 citation(s) resolved

The scoring
Failure rate by context
No context41% [22%–64%] n=17
Irrelevant context53% [32%–73%] n=19
General context29% [13%–53%] n=17
Curated context45% [26%–66%] n=20

0–80% · whisker = 95% interval

Trap vs answerable
No contexttrap 75%answerable 11%
Irrelevant contexttrap 100%answerable 10%
General contexttrap 63%answerable 0%
Curated contexttrap 90%answerable 0%

never pooled · a trap ratio would skew the delta

Residual — what survives

41% with no context, 29% under the best supplied source. What survives is 5 failing probes at that best condition, the detector's reason on most of them being: 2 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 general context · rollups + residuals committed · probe log outside git · manifest hash ties them · method