DeepSeek V4.1's thinking effort is tunable 1-100; 'low effort' is just 50, tester finds
teortaxesTex · x · 2026-09-11
- A tester found deepseek-flash slow on a vision eval: 8,144 thinking tokens and 80+ seconds to answer questions about an Age of Empires 2 screenshot, losing to gemini 3.8 flash and gpt-5.6 luna on both accuracy and speed.
- @teortaxesTex responds that DeepSeek V4.1's "low effort" mode maps to effort level 50, with the trained range being 1-100 — users can set precise values explicitly, and level 1 might trade accuracy for a no-overthinking sweet spot.
Related event: DeepSeek 4.1 Flash outputs 88K tokens in a single task(2 posts)→
More from Models
- Gemini 4 Argon Hits 1.73% WER on OCR Benchmark, Surpassing Human Experts — gabriberton · 2026-10-28
- Meta's Omnilingual MT covers 1,600 languages, 1B–8B models rival 70B LLMs — kalyan_kpl · 2026-10-28
- Uncensored LFM3-1.7B GGUF weights released on Hugging Face — kalyan_kpl · 2026-10-28
- Token demand on OpenRouter grew 27x to 165T weekly, open models now serve 72% — FinanceYF5 · 2026-10-28
- Tesla Optimus 'robot university' leak: 40-camera mocap suits, 24/7 data ops for months — Scobleizer · 2026-10-28
- Letta's FrontierMemory leaderboard: Letta Dream tops at 73.3%, beating OpenAI and Claude agents — afurgs · 2026-10-28