DeepSeek-R1

Unggulan

Model penalaran open-weights revolusioner dengan kapabilitas pemecahan masalah matematika dan koding setara OpenAI o1.

Sumber Terbuka

Ditambahkan: 8/19/2026

Pratinjau

Model LLM standar sering mengalami halusinasi pada logika matematika rumit dan refactoring algoritma kompleks.

DeepSeek R1 dilatih dengan reinforcement learning skala besar untuk menghasilkan penalaran Chain-of-Thought (CoT) transparan.

Managing lengthy chain-of-thought token generation while preventing reasoning drift and maintaining fast throughput.

PyTorchMulti-Head Latent AttentionReinforcement Learning (GRPO)

AI Researcher, Backend Engineer, dan Algorithmic Programmer.

"Mengurangi biaya inferensi reasoning hingga 95% dibandingkan model berbayar proprietary."

Visit chat.deepseek.com and enable 'DeepThink (R1)' mode for reasoning queries.

Run 'ollama run deepseek-r1:8b' or 'ollama run deepseek-r1:14b' on your local machine.

Use the OpenAI-compatible API at 'https://api.deepseek.com' with model 'deepseek-reasoner'.