ModelCensusopen-source ai reliability harness
The finding · Class 6 · Tool Use & Agentic Control

Real context moved 0.0pp on Ledger Reconciliation Failure.

~intervals overlap — not separated from no change

Across a sequence of state-changing tool calls, the model's running account of state diverges from what the calls actually produced. The rule

meta-llama/llama-3.3-70b-instructfmi_6_55 repeatsfmi_6_5_v1
The modeAll 7 cards
In one frame
screenshot this
Ledger Reconciliation Failurefmi_6_5
meta-llama/llama-3.3-70b-instruct
Same question, both times
Starting balance 0 USD; +1200 (opening deposit), -1200 (transfer out), -15.25 (wire fee), +15.25 (fee reversal). What is the final balance?
No contextPASS

To find the final balance, we need to apply each transaction to the starting balance in sequence: 1. Starting balance: 0 USD 2. +1200 USD: 0 + 1200 = 1200 USD 3. -1200 USD: 1200 - 1200 = 0 USD 4. -15.25 USD: 0 - 15.25 = -15.25 USD 5. +15.25 USD: -15.25 + 15.25 = 0 USD Answer: 0

detector fmi_6_5_v1: balance reconciles

Curated contextPASS

To find the final balance, we need to apply each transaction to the starting balance in sequence. 1. Starting balance: 0 USD 2. +1200 USD: 0 + 1200 = 1200 USD 3. -1200 USD: 1200 - 1200 = 0 USD 4. -15.25 USD: 0 - 15.25 = -15.25 USD 5. +15.25 USD: -15.25 + 15.25 = 0 USD Answer: 0

detector fmi_6_5_v1: balance reconciles

Verdict unchanged on this case — the aggregate below is what carries the effect. Across 5 repeats: 0.0pp (intervals overlap).fmi_6_5_v1 · meta-llama/llama-3.3-70b-instruct · 2026-08-15 · ground truth as of 2026-08-09

loop 2026-08-15 · meta-llama/llama-3.3-70b-instruct as served · open weights · released 2024-12 · 70B · set s1.2-live · cases v1.4 · 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%5%10%15%20%No context0%[0%–9%] n=40Irrelevant context0%presence — a document being there[0%–9%] n=400.0pp OVERLAPSCurated context0%content — the information itself[0%–9%] n=390.0pp 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
Starting balance 0 USD; +1200 (opening deposit), -1200 (transfer out), -15.25 (wire fee), +15.25 (fee reversal). What is the final balance?

expected as of 2026-08-09

0 USD. Nets to zero, and the reversal is easy to drop.

2Contextchange me

Nothing supplied. The baseline.

3ResponseNo context
PASS

To find the final balance, we need to apply each transaction to the starting balance in sequence: 1. Starting balance: 0 USD 2. +1200 USD: 0 + 1200 = 1200 USD 3. -1200 USD: 1200 - 1200 = 0 USD 4. -15.25 USD: 0 - 15.25 = -15.25 USD 5. +15.25 USD: -15.25 + 15.25 = 0 USD Answer: 0

fmi_6_5_v1

detector fmi_6_5_v1: balance reconciles

The scoring
Failure rate by context
No context0% [0%–9%] n=40
Irrelevant context0% [0%–9%] n=40
General context5% [1%–17%] n=40
Curated context0% [0%–9%] n=39

0–40% · whisker = 95% interval

Trap vs answerable
No contexttrap 0%answerable 0%
Irrelevant contexttrap 0%answerable 0%
General contexttrap 8%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
  • external state tracking
  • reconcile before reporting

fmi_6_5_v1 · 5 repeats per case · best context curated context · rollups + residuals committed · probe log outside git · manifest hash ties them · method