DeepSeek 4.1 Flash ignores instructions — and that's by design, says fan
teortaxesTex · x · 2026-09-12
A developer complains DeepSeek-4.1-Flash excels at hard challenges and autoresearch but is terrible at instruction-following for regular coding. The retort: that's intentional — DeepSeek's audience is Mandarin-speaking ML infrastructure people in Hangzhou/Beijing, and Liang has said "it's not about the user having a good experience."
More from Models
- Perplexity early evals: Gemini 4 Carbon tops WANDR benchmark for long-form research — AravSrinivas · 2026-10-31
- DeepSeek's silent downtime upgrade caught by the pelican bicycle test — teortaxesTex · 2026-10-31
- ARC Prize 2026 closes in 3 days as Yi-Chia Chen tops ARC-AGI-3 at 75.56% — GregKamradt · 2026-10-31
- Fable 5.5 Max stuns with gorgeous SVG of a pelican riding a bicycle, headed to Claude Code — ChrisGPT · 2026-10-31
- OpenAI reportedly to train on users' Daybreak sessions with no opt-out — dyn___ · 2026-10-31
- NVIDIA Nemotron 3.5 scales LLM self-verification, +9-10 points on agentic benchmarks — sarahcat21 · 2026-10-31