Coding-specialized DeepSeek V4 Flash hurts long-context novel and RP users
teortaxesTex · x · 2026-09-16
Chinese forum users report that the new DeepSeek V4.1 Flash is very poor at long-context fiction writing, matching earlier observations about V4 Flash's weak long-context chat. TeortaxesTex quips that this explains the uproar: the coding-specialized model is rough on novel and RP users, adding that web novel culture produced "poisonous training data" even before LLMs.
More from Models
- Open-source VoxLock tops Krisp voice isolation benchmark with 4.95% WER — Arbaz Siddiqui · 2026-11-02
- Anthropic's 85% inference margin sparks debate on open source's price-pressuring role — code_star · 2026-11-02
- User reports xAI appears to have quietly raised weekly usage limits, dropping his meter from 93% to 47% — XFreeze · 2026-11-02
- Meituan's LongCat-2.5 lands as 1.6T MoE agent model, with DeepSeek copycat speculation — teortaxesTex · 2026-11-02
- Asked what AI firms want you to stop owning, Claude said: your own compute — alex_verem · 2026-11-02
- Is Hermes Still Worth It? Codex's $200 Tier Now Overlaps Most of Its Value — Foreign_Lead_3582 · 2026-11-02