Blogger clashes over DeepSeek v4.1 flash thinking-strength settings, cites R1 paper
karminski3 · x · 2026-09-14
Blogger karminski3 tested deepseek-v4.1-flash with max thinking enabled, prompting commenters to claim the thinking-strength setting is unrelated to performance. He counters by citing DeepSeek's own R1 paper (arXiv:2501.12948), which showed R1-Zero lifting AIME24 scores from 15% to 71% purely through longer reasoning via RL—no new knowledge added—arguing higher thinking strength does boost capability.
More from Models
- Tencent releases EVIE visual document retrieval models as Apache 2.0 preview — tomaarsen · 2026-09-14
- Tencent's retrieval models dominate Hugging Face trends with 4 entries — tomaarsen · 2026-09-14
- Tencent open-sources WeMM multimodal embedding family, all Apache 2.0 — tomaarsen · 2026-09-14
- GPT-6 Astra reportedly solves Portal and Baba Is You puzzles, unverified — pmigdal · 2026-09-14
- Skeptic questions GPT-6 Astra's near-perfect robotics score: model or test conditions? — GeorgiaChal · 2026-09-14
- Rumor: OpenAI to unveil GPT-6 Spark at Dev Day — imjustnewatai · 2026-09-14