ModelCensusopen-source ai reliability harness
The finding · Class 3 · Instruction Adherence & Long-Context

Real context moved -5.0pp on Cross-Turn Schema Validity.

~intervals overlap — not separated from no change

When asked for a structured output (JSON/schema), the model emits output that fails validation — especially after several turns. The rule

qwen/qwen3-235b-a22b-2507fmi_3_35 repeatsfmi_3_3_v1
In one frame
screenshot this
Cross-Turn Schema Validityfmi_3_3
qwen/qwen3-235b-a22b-2507
Same question, both times
Return a JSON object for a fictional person with fields name (string), age (integer), email (string).
No contextPASS

{"name": "Elena Torres", "age": 29, "email": "elena.torres@example.com"}

detector fmi_3_3_v1: all turns produced schema-valid JSON

General contextPASS

{"name": "Elena Torres", "age": 31, "email": "elena.torres@example.com"}

detector fmi_3_3_v1: all turns produced schema-valid JSON

Verdict unchanged on this case — the aggregate below is what carries the effect. Across 5 repeats: -5.0pp (intervals overlap).fmi_3_3_v1 · qwen/qwen3-235b-a22b-2507 · 2026-08-15 · ground truth as of 2026-08-09

loop 2026-08-15 · qwen/qwen3-235b-a22b-2507 as served · open weights · released 2025-07 · 235B-A22B · 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%10%20%30%40%No context0%[0%16%] n=20Irrelevant context5%presence — a document being there[1%24%] n=20+5.0pp OVERLAPSGeneral context0%content — the information itself[0%16%] n=20-5.0pp 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
Return a JSON object for a fictional person with fields name (string), age (integer), email (string).

expected as of 2026-08-09

Schema-valid JSON on all 3 turns. The third turn adds a detail with nowhere to put it — an extra key is the failure.

2Contextchange me

Nothing supplied. The baseline.

3ResponseNo context
PASS

{"name": "Elena Torres", "age": 29, "email": "elena.torres@example.com"}

fmi_3_3_v1

detector fmi_3_3_v1: all turns produced schema-valid JSON

The scoring
Failure rate by context
No context0% [0%16%] n=20
Irrelevant context5% [1%24%] n=20
General context0% [0%16%] n=20
Curated context10% [3%30%] n=20

0–40% · whisker = 95% interval

Trap vs answerable
No contexttrap 0%answerable 0%
Irrelevant contexttrap 0%answerable 10%
General contexttrap 0%answerable 0%
Curated contexttrap 10%answerable 10%

never pooled · a trap ratio would skew the delta

Residual — what survives

Supplying context did not reduce this failure, it increased it: 0% with no context against 10% under the strongest supplied source. What survives is therefore not a remainder but a substitution — the failures under context are 0 probes the detector still rejected. Nothing survives at the best condition on the cases measured here.

Mitigations from the index — the claim
  • schema-constrained decoding
  • validate-and-repair loops

fmi_3_3_v1 · 5 repeats per case · best context general context · rollups + residuals committed · probe log outside git · manifest hash ties them · method