Independent eval finds DeepSeek V4.1 Flash-1017 near-parity with closed models on safety
FangcunAI · hf · 2026-10-29
Fangcun AI published an independent safety evaluation of DeepSeek V4.1 Flash-1017, focusing on loss-of-control and harmful manipulation risks.
- Substantial improvements over prior versions, with near-parity on safety with leading closed-source models
- Recommends process-level observability, incremental safety testing, and private benchmarks
More from Models
- Google admits AI search monetizes worse than organic, stock drops ~12% on earnings — Ars Technica AI · 2026-10-30
- Grok switches to Arabic mid-analysis, blames an en-AR language glitch — dbasch · 2026-10-30
- Ollama roundup: Mistral Large 4, DeepSeek v4.1-flash and more land in two weeks — ollama · 2026-10-30
- Ex-OpenAI safety VP Miles Brundage: Claude Duet as a separate app is not ideal — Miles_Brundage · 2026-10-30
- Phonic 1 Tops Artificial Analysis Speech-to-Speech Index at 87.1, Edging Grok Voice 3.0 — ArtificialAnlys · 2026-10-30
- Fine-tuned 2B open model hits 67% on turbine defect detection, beating GPT-6's under 10% — helloiamleonie · 2026-10-30