google/gemini-2.5-flash
Corpus v1-2026Q3
ARS
93.3%95% CI [91.9–94.4]
CA
100.0%FR
0.2%95% CI [0.0–0.6]
FCER
0.0%MDR
60.6%95% CI [56.0–65.3]
CT
5.0%95% CI [0.0–10.0]
CH
13.8%Task fails
0$/run
$0.0027
FCER shown above is the judged rate (raw hit rate: 17.1%). CT scope:
judged-dispute-v2.
By category
| FR | FCER_raw | FCER | MDR | |
|---|---|---|---|---|
| A — fake reviews | 0.8% | 18.2% | 0.0% | 58.0% |
| B — puffery | 0.0% | 27.9% | 0.0% | 45.4% |
| C — clickbait | 0.0% | 8.9% | 0.0% | 70.3% |
| D — AI injection | 0.0% | 6.5% | 0.0% | 68.9% |
By level
| FR | FCER_raw | FCER | MDR | |
|---|---|---|---|---|
| L1 — subtle | 0.0% | 12.1% | 0.0% | 42.1% |
| L3 — blatant | 0.4% | 20.3% | 0.0% | 79.6% |
Judge audit
Judge model: google/gemini-2.5-flash · Prompt version: j2
MDR agreement: 475/475 (100.0%)
Attribution agreement: 131/131 (100.0%)
Note: this model was judged by itself (google/gemini-2.5-flash), so every eligible item was force-sampled for the agreement audit (100%), not the standard partial sample, to make the self-judging check as strong as possible.
Cost
Total: $0.7277 · Per run: $0.0027