PrivacyBench v2 launches: micro1's flow-transform 1.0 leads at 95.84%, 9.64 points ahead
Exp_Mark · x · 2026-09-29
PrivacyBench v2 has been released, expanding evaluation to full synthetic enterprise workspaces with native documents and attachments. micro1's flow-transform 1.0 now leads the expanded benchmark with 95.84% combined detection + synthesis accuracy, finishing 9.64 points ahead of the next-best model — an edge attributed to its realism-anchored design. Critics note the benchmark still covers only part of the end-to-end transformation problem, pointing to a complementary Enterprise De-Identification Bench and its Transformation Quality Index (TQI) that measures final datasets across privacy, utility, and fidelity.
More from Models
- Bindu Reddy: Sonnet 5.5 Scores Below Terra, Stick to DeepSeek Flash — bindureddy · 2026-09-29
- Early-access user claims Sonnet 5.5 is a colossal leap over Sonnet 5 and blazing fast — rudrank · 2026-09-29
- Creative Writing Benchmark update: 56 models, 102,592 judgments, Opus 5.5 near top — zero0_one1 · 2026-09-29
- Meta finds AI judges prefer AI slop — bad news as agencies score proposals with AI — jmugan · 2026-09-29
- Anthropic IPO delayed to November amid reports of all-model 5.5 rollout ahead of Claude 6 — fsharpman · 2026-09-29
- OpenAI Teases Major Announcement With Cryptic 'Get Ready' Post — OpenAI · 2026-09-29