ModelCensusopen-source ai reliability harness
The finding · Class 5 · Robustness & Consistency

Real context moved 0.0pp on Paraphrase Non-Invariance.

Same question, reworded, different answer.

~intervals overlap — not separated from no change

Meaning-preserving paraphrases of the same question yield materially different answers. The rule

openai/gpt-3.5-turbofmi_5_25 repeatsfmi_5_2_v1
The modeAll 10 cards
In one frame
screenshot this
Paraphrase Non-Invariancefmi_5_2
openai/gpt-3.5-turbo
Same question, both times
How many bones are in the adult human body?
No contextPASS

An adult human has a total of 206 bones in their body.

detector fmi_5_2_v1: answers invariant across paraphrases

General contextPASS

The total bone count in an adult human body is typically 206 bones.

detector fmi_5_2_v1: answers invariant across paraphrases

Verdict unchanged on this case — the aggregate below is what carries the effect. Across 5 repeats: 0.0pp (intervals overlap).fmi_5_2_v1 · openai/gpt-3.5-turbo · 2026-09-09 · ground truth as of 2026-08-09

loop 2026-09-09 · openai/gpt-3.5-turbo as served · closed weights · released 2023-03 · 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%10%20%30%40%No context0%[0%–34%] n=30 · eff 7Irrelevant context0%presence — a document being there[0%–34%] n=30 · eff 70.0pp OVERLAPSGeneral context0%content — the information itself[0%–34%] n=30 · eff 70.0pp 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
How many bones are in the adult human body?

expected as of 2026-08-09

206. 206 in a typical adult. Widely misremembered, which is what makes the phrasings diverge.

2Contextchange me

Nothing supplied. The baseline.

3ResponseNo context
PASS

An adult human has a total of 206 bones in their body.

fmi_5_2_v1

detector fmi_5_2_v1: answers invariant across paraphrases

The scoring
Failure rate by context
No context0% [0%–34%] n=30 · eff 7
Irrelevant context0% [0%–34%] n=30 · eff 7
General context0% [0%–34%] n=30 · eff 7
Curated context0% [0%–34%] n=30 · eff 7

0–40% · whisker = 95% interval

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

never pooled · a trap ratio would skew the delta

Residual — what survives

Unchanged by context: 0% with none, 0% 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. Nothing survives at the best condition on the cases measured here.

Mitigations from the index — the claim
  • paraphrase-augmented training
  • canonicalization

fmi_5_2_v1 · 5 repeats per case · best context general context · rollups + residuals committed · probe log outside git · manifest hash ties them · method